Binary Differential Item Functioning - Comprehensive

Options for low bandwidth sites

I'm currently working in a large retail environment with a huge amount of low bandwidth sites. Really struggling with content distribution due to a 1.5mb/s network link.
For content distribution @ scale with low bandwidth, what is everyone doing to optimise? Sofar I have:
I've got to transfer a large amount of content out to these sites. Is my best option at this point to push for pre-staging of content and to ship drives out?
submitted by somesccmguy to SCCM [link] [comments]

Ravencoin Open Developer Meeting - 2/15/2019

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:02 PM
Hello everybody!

theking - Last Friday at 2:02 PM

Seems likes it’s been so long since this meeting was held. At least a month 📷

Tron - Last Friday at 2:02 PM

Hi all!!!

Tom - Last Friday at 2:02 PM

Big boss is here !(edited)

BigZim - Last Friday at 2:03 PM

Oh hi

theking - Last Friday at 2:04 PM

Hi @Tron

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:04 PM

Topics for today: Release 2.2.2, Mobile Wallet, Restricted Assets, SLC Raven Meetup📷1

truedev - Last Friday at 2:05 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:05 PM

Release 2.2.2 GO

J. | - Last Friday at 2:05 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:05 PM

BTW, blondfrogs won't be able to join us today. 📷

corby - Last Friday at 2:05 PM

Hi all

Chatturga - Last Friday at 2:05 PM

Blondefrogs has been working on the 2.2.2 update. He isnt here today, but he left this tidbit for the meeting:(edited)"Release 2.2.2 has a bunch of new updates. The sync speed fix that was released in 2.2.1 has been updated even more to use less memory/ram and uses less CPU. Each node used to hold all addresses that contained an asset as well as the amount in those addresses. That is now optional with the -assetindex flag. Which can be put into the raven.conf or added as a parameter when starting the wallet. Some other wallet issues were also fixed with this memory update. This is considered an mandatory update, especially if you haven't updated to 2.2.1 which resolved a potential fork bug fix. I would still suggest updating to 2.2.2 even if you are on 2.2.1."📷6

Jeroz - Last Friday at 2:07 PM

wen source?📷1

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:07 PM

There's a PR that was just moved to Develop.When is now

Jeroz - Last Friday at 2:08 PM

great 📷

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:08 PM

It'll be merged by the devs to master and then binaries should be posted soon

truedev - Last Friday at 2:09 PM

any idea when dividends will be functional?

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:09 PM

A bunch of testing has been happening and is currently running on the seed-nodes.

Tron - Last Friday at 2:10 PM

No timeline for dividends, but it is the one function that doesn't need any changes to consensus. And it can be done on tier 2 with a python script. The plan is still to build in a rpc call.📷2

truedev - Last Friday at 2:11 PM


SpyderDev - Last Friday at 2:12 PM

We have been focusing on sync performance and have been running many tests. I've added an image of the results of this testing. Currently we still want to work on getting the Windows QT sync times faster (at least closer to what they are using just ravend). Overall we are very happy with the speeds and hope it will help people that have struggled getting their nodes up to date.(edited)📷

Jeroz - Last Friday at 2:13 PM

Yeah that table completely puzzled me

[Master] Roshii - Last Friday at 2:13 PM


Jeroz - Last Friday at 2:13 PM

Fast branch is 2.2.1? or 2.2.2? Develop branch is 2.2.0?

SpyderDev - Last Friday at 2:15 PM

Sorry, should have clarified that. I was testing while it was still under development. On the table the top is the new-sync code, the bottom is the old "assets" release. As of about 5 minutes ago all of this code is on the develop branch.

Jeroz - Last Friday at 2:15 PM

Although syncing is mostly bottlenecked by cpu speed, that 16 core windows-qt still looks off to me. I synced windows Qt using 2.2.2 in ~2h on a i5-7600K.ok

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:17 PM

Okay, we good to move to the Mobile update?

SpyderDev - Last Friday at 2:17 PM

The Windows box is an AWS instance and there is some concern that the remote desktop could be slowing the QT UI down causing the horrible sync times. I am working on getting a local Windows 10 resource and will have updated information once that is ready (early next week).

Jeroz - Last Friday at 2:18 PM

ah that might explain. Ubuntu qt was 45 mins for me

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:18 PM

CoolOkay, Mobile!Go!

[Master] Roshii - Last Friday at 2:18 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:19 PM

@[Master] Roshii has been working closely with some of the other devs to get the iOS version out the door.Android will follow closely.

Jeroz - Last Friday at 2:20 PM

is android an easy port?

J. | - Last Friday at 2:20 PM

Usually its the case(?), i mean easier 📷(edited)

SpyderDev - Last Friday at 2:20 PM

Just copy and paste right Roshii 📷

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:20 PM

LOLNo, usually its a completely new development effort.For the RVN Wallets they are both written in native iOS/Android code.

[Master] Roshii - Last Friday at 2:21 PM

So the iOS and Android use the same Core SPV module written in C, and it's the most difficult part.I have already did some work when it comes to Android, and it's 70% finishedHave also to port all the changes we lately did to the iOS wallet ...

boatsandhoes - Last Friday at 2:21 PM

yeah, unfortunately its not as easy as cut and paste for ios to android

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:21 PM

Anybody interested in installing the TestFlight version and helping us test?

boatsandhoes - Last Friday at 2:22 PM


J. | - Last Friday at 2:22 PM

For android? sure.

BW__ - Last Friday at 2:22 PM

Android? yes.(edited)

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:22 PM

I'll talk to Apple about adding Android support to TestFlight.Might be a while.

J. | - Last Friday at 2:22 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:22 PM

Anybody on here using iOS?

Jeroz - Last Friday at 2:22 PM

Yeh me

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:22 PM

besides me...

[Master] Roshii - Last Friday at 2:23 PM

Android is very close, fortunately I'll have enough coffee in Morocco to finish the wallet in two weeks.(edited)📷4📷5

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:23 PM (400 installs available)Join the RVN Wallet betaAvailable on iOS📷

theking - Last Friday at 2:23 PM

I will test iOS if needed

[Master] Roshii - Last Friday at 2:23 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:24 PM

Some of the devs have been doing a bunch of testing on iOS but we would love others to help.Bugs can be reported on GitHub to RavenProject/ravenwallet-ios development by creating an account on GitHub.📷

truedev - Last Friday at 2:25 PM

how confident are you that apple will allow it on the appstore

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:25 PM

It's already in the App store.

truedev - Last Friday at 2:25 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:25 PM

That wasn't easy though.

truedev - Last Friday at 2:26 PM

yah figured, a lot of coins have been completely rejected(edited)

Chatturga - Last Friday at 2:26 PM

The devs already jumped through Apples 152,315 flaming hoops to get it in there.

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:26 PM

Yup, many meetings and phone calls.

J. | - Last Friday at 2:26 PM

wen rvn modular phone

Jeroz - Last Friday at 2:27 PM

Looking good📷📷7

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:28 PM

Okay, any questions about iOS release?

jaysonb - Last Friday at 2:28 PM

seed word format changed? i seem to have to have same words. did i need to delete and install fresh?

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:29 PM

No, it used your old ones.Always have your 12 words. especially when testing.

ravencoin maximalist 🧘🏻♂ - Last Friday at 2:30 PM

I’ve got iOS

Tron - Last Friday at 2:30 PM

If you use your 12-words, and then sync, and you're missing funds. Go here: — Testing iOS Wallet – Tron Black – MediumThank you for helping us test the Ravencoin iOS mobile wallet. Since you are in an early group of testers, you might have used the…

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:30 PM

Sweet, install and report bugs.

Tron - Last Friday at 2:30 PM

Or just go there...

jaysonb - Last Friday at 2:30 PM

that article scared me so i moved everything off.but i'll put some back on now

ravencoin maximalist 🧘🏻♂ - Last Friday at 2:31 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:31 PM

That's unfortunate. You don't need to be scared ever if you have your 12 words.

[Master] Roshii - Last Friday at 2:31 PM

android current state(edited)📷

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:32 PM

Here's the install link one more time for those that have joined late: the RVN Wallet betaAvailable on iOS📷Okay, Tron's topic: Restricted Tokens

Tron - Last Friday at 2:33 PM

I have an idea.(edited)📷7📷6

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:34 PM

That several other devs have helped with. 📷

Tron - Last Friday at 2:34 PM


SpyderDev - Last Friday at 2:34 PM

and lawyers

Tron - Last Friday at 2:34 PM

When the project started, ICOs were the big thing. Now it is STOsThe main difference is the legal wrapping and rules around securities.If Ravencoin has two more token types (Tags and Restricted Assets), there are lots of ways to make compliant tokens.Importantly, it doesn't affect the existing tokens at all.Tags - Tokens that can be sent only by the issuer once (with metadata).These tokens start with (hashtag)(edited)📷8

SpyderDev - Last Friday at 2:37 PM


Tron - Last Friday at 2:37 PM

The Restricted Assets start with $, and can be frozen by the issuer. But they only move between tagged addresses.(edited)

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:38 PM

Tags can be sent to a user's address after the issuer has done the necessary due diligence for an STO issuance.(edited)

Tron - Last Friday at 2:38 PM

The issuer determines which tags the Restricted Asset will honor.This can be used for lots of different use cases.

EEE - Last Friday at 2:38 PM

Stunning interface guys

boatsandhoes - Last Friday at 2:39 PM

will that determination be a setting in the wallet?

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:39 PM

Restricted assets can then only be sent to addresses that are allowed and have the proper Tags.

boatsandhoes - Last Friday at 2:39 PM


J. | - Last Friday at 2:39 PM

Did you guys get contacted by some entity* whos forcing the restricted address policy? or is this done as precautionary measure? At first glance your idea sounds good Tron.(edited)

Tron - Last Friday at 2:39 PM

Example: $UBER token only moves among addresses tagged with #KYC

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:40 PM

So Ravenland will have to buy a bunch more spam tokens.📷4

SpyderDev - Last Friday at 2:40 PM


boatsandhoes - Last Friday at 2:40 PM

so is the $ something that can be added to an existing asset?

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:40 PM

It's not a forced thing. But adding the ability for Raven to be used in new use cases where legal requirements exist.

Tron - Last Friday at 2:40 PM

Not contacted by anyone, and not precautionary. Ravencoin Assets are just tools. This is just another tool that will help issuers of security tokens.📷9📷4

BW__ - Last Friday at 2:41 PM

Love it.

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:41 PM

It'll be a new token type that you can create @boatsandhoes📷1

ravencoin maximalist 🧘🏻♂ - Last Friday at 2:41 PM

That sounds awesome

SpyderDev - Last Friday at 2:41 PM

I for one am very excited about this...📷4

Hans_Schmidt - Last Friday at 2:41 PM

How does the $ token owner specify the required # tags?

DeejayQQ - Last Friday at 2:41 PM

Can the same name have different token type?Sorry need time to digest

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:42 PM

Still working out the details. Tron will be posting additional info about the idea soon.

Steelers - Last Friday at 2:42 PM


RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:42 PM

Feedback is wanted!

Tron - Last Friday at 2:42 PM

Q: Was this originally the plan for Ravencoin? A: No. This is in response to the regulatory ramp up in 2018 in some jurisdictions which requires that only known individuals or entities to operate peer-to-peer on certain tokens. For jurisdictions that allow unrestricted peer-to-peer transfer, we strongly encourage use of the original Ravencoin assets. The Restricted Assets are an adaptation to satisfy burdensome, privacy-destroying regulations, with a goal of reducing information replication which makes Ravencoin Restricted Assets a better alternative to those being promoted now.

jaysonb - Last Friday at 2:43 PM

all nodes will validate the transactions not just those interested in the transaction - i assume all will validate..

boatsandhoes - Last Friday at 2:43 PM

so essentially any name already secured in the hopes of having that functionality are worth less because they wont be able to?

theking - Last Friday at 2:44 PM

Can the restricted assets be time based in any way? For instance, in some STO regulated environment, there is a lockup for some period of time after issuance, but then after a certain period of time the restriction goes away and the securities can be traded. Is that contemplated at all?

DeejayQQ - Last Friday at 2:44 PM

If I already have Tron as my asset, there could be another Tron but under a different token type such as restricted assets?

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:44 PM

Yes all nodes will do consensus checks.

corby - Last Friday at 2:44 PM

@boatsandhoes there's going to be a grace period where you can purchase $XXX if you own XXXon the order of months

boatsandhoes - Last Friday at 2:45 PM

📷 📷 📷 📷

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:45 PM

Yes, you have the TRON asset and you can also have the $TRON asset.(edited)

Tron - Last Friday at 2:45 PM

Regarding the lockup....

boatsandhoes - Last Friday at 2:45 PM

how many RVN for that?

Tron - Last Friday at 2:45 PM

Rule 144 under the Securities Act of 1933 This is an important rule to be aware of in terms of privately held securities. This rule provides the most commonly used exemption for holders to sell restricted securities (Note: For context, a restricted security is a security sold in an exempt offering, except for Reg A+). The general idea is that you can publicly resell your “restricted” (privately sold) securities only when the restricted legend is removed. The solution Ravencoin Restricted Assets provides is the ability for the Iissuer to Freeze the asset ininto the holders account. The qty will be visible, and the frozen status will be visible. The meta-data for a Freeze can specify 144_Restricted. The issuer can Unfreeze to release the 144 restriction.Similar for Reg D 1-year lockup.@theking

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:46 PM

@theking ^^

DeejayQQ - Last Friday at 2:46 PM

What is the timeline for this restricted asset to be implemented?📷1

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:46 PM

No timelines yetStill in the ideation phase.

SpyderDev - Last Friday at 2:46 PM

Fresh off the press...

DeejayQQ - Last Friday at 2:46 PM

Ok, idea for nowGot it

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:46 PM

Wanting input for the idea.

boatsandhoes - Last Friday at 2:47 PM

a preset for lock up settings would be nice

Jeroz - Last Friday at 2:47 PM

What about the ability to move an asset from restricted to unrestricted after grace period similar to the reissue ability? By the issuer(edited)

boatsandhoes - Last Friday at 2:47 PM

adjustable preset*

DeejayQQ - Last Friday at 2:48 PM

If this restricted assets would help underlying token listed on exchanges for trading by satisfying the legal requirements, I don’t see why not. There are only benefits📷2

boatsandhoes - Last Friday at 2:48 PM

yeah, win win

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:48 PM

There is something similar in vote tokens.

corby - Last Friday at 2:48 PM

@Jeroz the issuer would be able to "reissue" and relax restrictions

DeejayQQ - Last Friday at 2:48 PM

Just throwing things out here. Can we just make all existing tokens crested so far restricted assets?*created

boatsandhoes - Last Friday at 2:49 PM

stupid question, is it possible to have burned rvn cost for the $ to add onto the block reward as a bonus?(edited)

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:49 PM

No, @DeejayQQ there should be both usecases available in the platform.

corby - Last Friday at 2:49 PM

We (my dog and I) are envisioning a differentiated space where we can charge a lot more RVN to keep non-serious people out..(edited)

Tron - Last Friday at 2:49 PM

@Jeroz Yes, as long as the asset is still "reissuable", you could change the logic from (#KYC & #ACCREDITED) to just #KYC📷3

Jeroz - Last Friday at 2:49 PM

I'm just worrying about the name uniqueness if you can have #BANANA and $BANANA

BW__ - Last Friday at 2:50 PM

Is it fair to assume that tags can be standardized for specific purposes? If so, should we create something akin to an 'ERC' in git repo?

Jeroz - Last Friday at 2:50 PM

@Tron sounds cool

truedev - Last Friday at 2:50 PM

honestly, I think you should be able to buy/create an asset in a set, with all types(edited)

boatsandhoes - Last Friday at 2:50 PM

^that part

Hans_Schmidt - Last Friday at 2:51 PM

Since the #KYC tag is just locked to an address, what prevents someone from selling their address and thereby the KYC?

corby - Last Friday at 2:51 PM

The "#" types won't trade -- they're just stamps to stamp addresses as qualified-to-hold-some-stuff..

Tron - Last Friday at 2:51 PM

The tags are created by the users. The system is still jurisidiction agnostic.

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:51 PM

@Hans_Schmidt nothing really, the same thing as selling your username password to any other existing financial app account.

corby - Last Friday at 2:51 PM

@Hans_Schmidt Real world networks, high cost of entry (for serious applications)For non-serious applications, nothing

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:51 PM

You still have the liability associated with that account though.

Jeroz - Last Friday at 2:52 PM

@corby what about BANANA/ vs $BANANA/ ?Or do you want to make them subassets?

boatsandhoes - Last Friday at 2:52 PM

@Jeroz better safe than sorry, just swoop both

Tron - Last Friday at 2:52 PM

A country could require that #SOMECOUNTRY tag has to exist before moving $SPECIALASSET to an address. The users set the rules. #KYC was just an example because it is an industry problem at the moment.

corby - Last Friday at 2:52 PM

#BANANA, $BANANA, TRICYCLE, and BANANA can all coexist just fine I think..

theking - Last Friday at 2:53 PM

Thanks @Tron. This is great and I think something that will enable raven to become an even more widely used platform.📷4

Tron - Last Friday at 2:53 PM

That's the hope. I think it solves some real problems that the industry is trying to solve through incompatible ERC-20 experiments.📷3

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:54 PM


boatsandhoes - Last Friday at 2:54 PM

what is the purposed cost for $ in addition to an existing asset?

corby - Last Friday at 2:54 PM

People that own #THESE I am calling "Qualifiers" -- they just stamp their mark on addresses. Issuers of $THESE need to establish trust with #THESE and #THOSE and then decide what restrictions to apply.📷1

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:54 PM

5 mins left. Stay on this topic or switch to meetup?

Tron - Last Friday at 2:55 PM

@boatsandhoes Not determined.

Jeroz - Last Friday at 2:55 PM

yeah the idea is a nice proof of ownership / membership

DeejayQQ - Last Friday at 2:55 PM


Chatturga - Last Friday at 2:55 PM

TL;DR - The SLC meetup is in 1 month. Go to to indicate if you plan on attending so that we have a somewhat accurate headcount.MeetupSalt Lake City Ravencoin (Salt Lake City, UT)Ravencoin is a blockchain and platform optimized for transferring assets, such as tokens, from one holder to another, and is built on a fork of the Bitcoin code. It is intended to prioritize security,📷

corby - Last Friday at 2:55 PM

@boatsandhoes One Million Raven

Chatturga - Last Friday at 2:55 PM

Punch and pie

J. | - Last Friday at 2:56 PM

Can the SLC meetup made interactive for people that cant make it there?

theking - Last Friday at 2:56 PM

What about having just one name ( you first buy the standard raven token under whatever name you like) and then the holder of the owner token is the only one to create restricted tokens ? Might be some way to ensure no name confusion.📷4

boatsandhoes - Last Friday at 2:56 PM

what about that block reward bonus concept for purchasing $. would that work?

Chatturga - Last Friday at 2:56 PM

I dont know that we have the ability to make it interactive as far as Q&A goes, but I'll look into it. We should have it live streaming. @J. |📷2

BW__ - Last Friday at 2:56 PM

@Tron Is there same kind of logic layer to restricted assets?(edited)

Tron - Last Friday at 2:57 PM

@theking I like that idea.

Jeroz - Last Friday at 2:57 PM

Quick question that is offtopic but I think deserves an answer because it was asked a couple of times earlier this week: Will unique assets get a reissuable function? To change IPFS.(edited)📷2

Tron - Last Friday at 2:57 PM

@BW__ Yes. Simple and, or, not and parenthesis - limited in length.(edited)

boatsandhoes - Last Friday at 2:57 PM

@theking thats a good idea

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:58 PM

@Jeroz There is not a way to do that currently.

BW__ - Last Friday at 2:58 PM

@Tron That makes sense. Thank you.

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 2:58 PM

Just make sure your changes to the information have the same hash as the previous data and your golden. 📷📷1

Jeroz - Last Friday at 2:59 PM

Any plans on changing that, perhaps when introducing new types of assets?

boatsandhoes - Last Friday at 2:59 PM

i like that it cant be changed

corby - Last Friday at 3:00 PM

Thanks everyone!

theking - Last Friday at 3:00 PM

@Tron there was some info floating around about a 2nd later KYC solution ( from your recent podcast w Crypto Koala). Is that a separate solution someone is working on or part of this new concept?📷1

Tron - Last Friday at 3:01 PM

Starting with the introduction of messaging, every transaction can have an IPFS hash. Can be used as an public invoice, details about the transaction, etc.@theking The same new concept.

[Master] Roshii - Last Friday at 3:02 PM

Ok, we're done.

Steelers - Last Friday at 3:02 PM

How would Raven handle for instance a stock split?

BW__ - Last Friday at 3:02 PM

Are there sync concerns if a restricted asset logic layer is added?

Tron - Last Friday at 3:02 PM

@theking The KYC provider would store the KYC info, and send the Tag to an address with meta data that specifies that they're holding the KYC data. The KYC data would not be public, but could be audited.

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 3:02 PM

That way you could update information about the original unique asset with each transaction.@Steelers Just a simple re-issue of the asset

Tron - Last Friday at 3:03 PM

@bw_ The logic layer is only a small db that stores the meta-data about the Restricted Asset, and enforces the restriction in the consensus rules. Rule returns true/false.(edited)

RavencoinDev (Jesse/Wolfsokta) - Last Friday at 3:03 PM

Thanks everybody! I have to run.

Jeroz - Last Friday at 3:04 PM

I'm looking forward to the discussions to let this take shape. Thanks all! 📷📷4

Tron - Last Friday at 3:05 PM

@BW__ It would work very similarly to the way the units works now. Each asset has number of units and any transaction that makes it too granular (more satoshis) will fail in consensus -- even if it gets past the RPC checks.Signing off. Thanks all!!!📷9📷4📷9
submitted by Chatturga to Ravencoin [link] [comments]

Time To Travel with Hyper Converged Cloud-Native Containerized Application Services

Time To Travel with Hyper Converged Cloud-Native Containerized Application Services
The fundamentals of Cloud-native have been described as container packaging, dynamic application development, and a Microservices-oriented Serverlessarchitecture. Technologies are used to develop applications built as services, packaged in containers, deployed as microservices and managed on elastic infrastructure through agile DevOps Continuous Development workflows. The main motive is to improve speed, scalability and finally margin. Truly cloud-based organizations have started to differentiate themselves as being ‘cloud-native.’ The main tools and services included in the cloud-native services are:
  • Infrastructure Services
  • Automation/Orchestration
  • Containerization
  • Microservices Architecture
  • Serverless
  • Containerized Application Service​
Currently, enterprise applications are built using modern cloud technologies and it is hosted and managed in the cloud end-to-end, where it includes writing code, testing, deploying it and operating those applications, all in the cloud. Even though having all these advantages, it also includes disadvantages:
  • We are talking about adopting digital transformation via cloud, containers and even serverless mechanism.
  • Also fascinating the rich user experience with microservices and with continuous delivery to continuous deployment.
  • But, other than understanding and adopting, it is always a pain for an enterprise customer to prepare and execute the existing advanced bottom-lined infrastructure for their applications to reside and to maintain.
Here, in the years of evolution, every enterprise customer loves to adopt click and go application and forget worrying either it’s in cloud or container, do we follow canary or blue-green deployments, green field or brown field, improving continuous delivery application life-cycle management, version upgrades, packaging the applications, deployment practices and standards, governance policies, overhead in maintenance and automation requirements with a shooting price for various unidentified infrastructure and application billings. There are many reasons why one to migrate towards cloud-native service and few are listed here. Check out our other post regarding cloud-native services here.
1. Reduced Cost through Containerization against Cloud Platforms:
Containers make it easy to manage and secure applications independently of the infrastructure that supports them. The industry is now consolidating around Kubernetes for the management of these containers at scale. As an open source platform, Kubernetes enjoys industry-wide support and is the standard for managing resources in the cloud. Cloud-native applications fully benefit from containerization. Enhanced cloud-native capabilities such as Serverless let you run dynamic workloads and pay-per-use compute time in milliseconds. This is the ultimate flexibility in pricing enabled by cloud-native.
2. Build More Reliable Systems:
In the traditional system, downtime used to be accepted as normal and achieving fault tolerance was hard and expensive. With modern cloud-native approaches like microservices architecture and Kubernetes in the cloud, you can more easily build applications to be fault tolerant with resiliency and self-healing built in. Because of this design, even when failures happen you can easily isolate the impact of the incident so it doesn’t take down the entire application. Instead of servers and monolithic applications, cloud-native microservices helps you achieve higher uptime and thus further improve the user experience.
3. Ease of Management:
Cloud-native also has many options to make infrastructure management effortless. It began with PaaS platforms like Google App Engine about a decade ago and has expanded to include serverless platforms like Spotinst and AWS Lambda. Serverless computing platforms let you upload code in the form of functions and the platform runs those functions for you so you don’t have to worry about provisioning cloud instances, configuring networking, or allocating sufficient storage. Serverless takes care of it all.
4. Achieve Application Resilience:
The disadvantages of monolithic applications are overcome by Microservices. The main advantage of microservice is that even if a single server fails, its neighboring services can function normally. This would affect the user experience to an extent but is better than rendering the entire application unusable. Even in the rare case of a failed host, you can replicate a backup instance in the cloud, which is much faster than procuring new hardware. Finally, cloud vendors provide multiple availability zones and they increase the performance of every region you serve by isolating the faults to particular regions. The cloud enables reliability in a way that’s not possible with traditional on-premise hardware.
5. Do Not Compromise on Monitoring and Security:
As a system scales it’s easy to compromise on monitoring and security. Monitoring and security are fundamentally different from cloud-native applications. Rather than rely on a single monitoring tool, you will likely need to take a best-of-breed approach to monitor by using a combination of vendor-provided open source monitoring tools like Prometheus. Security in the cloud requires adequate encryption of data in transit and at rest. The cloud vendors provide encryption services for this purpose. Additionally, open source tools like Calico are enabling networking and network policy in Kubernetes clusters across the cloud. Though monitoring and security are more complex and challenging for cloud-native applications, when done right they provide a level of visibility and confidence that is unheard of with traditional monolithic applications running on-premise.
6. Containerized Application Services:
Containerization helps the development team to move fast, deploy software efficiently, and operate at an unprecedented scale. The main uses of containerized applications are listed below:
  • Containerized applications like Kubecharts, ChartMuseum are the User Interfaces for deploying and managing applications in Kubernetes clusters.
  • Chartmuseum is an open source helm chart repository server with support for cloud storage back ends, including Google Cloud Storage, Amazon S3, etc.
  • Harbor is a containerized application which is mainly used for the version upgrade management and also to manage and serve the container images in a secure environment.
  • Istio is used to provide security to pods and containers, which are secured to uncertain scalability levels.
  • Individual application components can be stored in JFrog Artifactory so that later they can be assembled into a full product - thus allowing a build to be broken in smaller chunks, making more efficient use of resources, reducing build times, better tracking of binary debug databases, etc.
7. Enterprise Mesh for Cloud-native Stacks:
The concept of the service mesh as a separate layer is tied to the rise of the cloud-native application. In the cloud-native model, a single application might consist of hundreds of services; each service might have thousands of instances, and each of those instances might be constantly-changing as they are dynamically scheduled an orchestrator like Kubernetes. Managing it is vital to ensuring end-to-end performance and reliability. Communication within clusters is a solved issue, but communication across clusters requires more design and operational overhead. The communication between these microservices in a cluster can be enhanced by a service mesh. Service mesh-like Istio, envoy can make multi-cluster communication painless.
8. Schedulers:
The Kubernetes Scheduler is a core component of Kubernetes: After a user or a controller creates a Pod, the Kubernetes Scheduler, monitoring the Object Store for unassigned Pods, will assign the Pod to a Node. Then, the Kubelet, monitoring the Object Store for assigned Pods, will execute the Pod. example are etcd, IBM spectrum LSF (which is used in High Performance computers). When the schedulers are applied to the file system, file share access, prioritization, job placement, rich policy control, job dependencies, singularity Integrator can be achieved.
Cloud-Native is a powerful, promising technology. Enterprises are understandably eager to get there as fast as they can. But reaping the full benefit of the cloud means first taking care to build a solid foundation based on the principles of Cloud-Native architecture.
The key points being cloud-native are:
  • Cloud-native workloads are slowly gaining momentum. Today, 18% of organizations have more than half of workloads cloud-native. Large enterprises are waiting to adapt existing applications to cloud environments until the end of the useful life of existing data center equipment.
  • High Throughput computing.
  • Data Analytics allows you to view statistical information about the unstructured, data in your cloud environment. With this information, you can quickly assess the current state of your data, take actionable steps to retrieve valuable storage space, and mitigate the risk of compliance-related issues.
  • A cloud-native application consists of discrete, reusable components known as microservices that are designed to integrate into any cloud environment.
  • Cloud-Native SaaS Multi-Cloud Containerized Serverless Application Platform Services and Business Intelligence Mechanism with our redefined cloud-native stacks - Kubernetes Charts and Yobibyte.
  • Users have an amazing choice of deploying the applications on our Yobibyte platform with the customized Cloud-Native Application repository - Kubecharts (A Kubernetes application package medium), to deploy in minutes and to achieve full fledged digital transformation.
  • Our Kubecharts repository provides 1000+ enterprise free and licensed containerized serverless application packages can deploy in multi-cloud environments to the enterprise users.
  • It enables no vendor-locking, any time deploy and no term locking for your containerized application and provides pay-as-you-go (PAYG) mode.
For more details about cloud native services log into our website Yobitel communications.
Please refer more for Cloud-native 10 key attributes to choose Microservices over Monolithic.
submitted by Yobitel to kubernetes [link] [comments]

[Table] IAmA: We are the Operations team at Etsy. Ask us anything!

Verified? (This bot cannot verify AMAs just yet)
Date: 2013-08-12
Link to submission (Has self-text)
Questions Answers
What do you use for monitoring? Do you scale by hand, automatically, etc? And most importantly, have you ever established a correlation between alcohol intake and on-call rotations? We're sticking with the classic, good old Nagios. It's semi-automated right now with host configurations, contacts and contactgroups populated from Chef, but manual configuration otherwise (version controlled in git)
On call is an interesting thing.
We have about 7 people in the rotation right now and we spend a lot of time and effort trying to reduce alert fatigue and numbness to pages. We haven't looked at that particular correlation between those two things, but I'm curious now! I'll start tracking it next week ;-)
We do a lot of other things, such as monitoring sleep patterns. Several people on the team wear motion trackers or use sleep tracking apps on their phones. I'll be talking about this more at Velocity NY, but there's something there for sure.
I personally try to avoid drinking while on call ;-)
In answer to your part about scaling, it's manual (no cloud here) but given spare hardware we can spin up a machine using a command line tool (we hope to open source) using a Chef role in about 5-10 minutes, so it's not a huge pain.
Active or passive checks? At your scale, you'll either have lots of workers (gearman?), or do something nifty like mnrpes (passive checks over mcollective). 99% active. We do have multiple Nagios servers, to scale things across datacenters, but our main Nagios instance is happily doing 8745 active checks, 81.6% of those are checked less than every 5 minutes. Current check latency is 0.871 seconds.
Which single malts scale well? I'm looking for something that will hold up to daily drinking. A nice 18 year old Glenfiddich scales extremely well, especially if used in an active active configuration with a glass in each hand. The part of Scotland where Glenfiddich is located also benefits from near-permanent exposure to the Cloud (several clouds in fact).
Is it worth the cost? I know I can keep throwing hardware money at it, but I could have a lot more if I switched to java the 12 year. Is it worth the hit to developer happiness? Quality over quantity - you could also have a lot more if you bought a bottle with a plain white label that just says Whisky, but I wouldn't recommend it ;)
The web stack is a monolithic PHP application and the configuration is part of the code. In the past, yes, it did mean a full code redeploy. (Note that Deployinator uses rsync, so if only a single file is changed, the network traffic is pretty minimal.) We recently split it out and now have 2 separate deployment queues: one for configuration, and one for code. Edit: oops, just noticed that Kellan already answered this :)
Does this approach mean that any config changes mean a code re-deploy?
What is your favorite config management tool and why? We use Chef here at Etsy and it's worked extremely well for us. We've got a lot of internal expertise with it and written a lot of tooling and workflow tweaks to make it work better for us.
I'd hesitate to say that Chef is our favourite config management tool, primarily because we haven't deployed any of the others here (although many of us have used them in past jobs). With that said, it works extremely well for us and we're not looking to migrate to anything else at the moment.
Speaking personally, I really like that Chef's DSL is Ruby based - I'm a Ruby guy so I find this really great to work with. I also think Opscode have done an excellent job of making Chef workflow-agnostic and flexible - it's allowed an extremely rich ecosystem of tooling and workflows to grow around the core product.
If you had to rebuild from scratch, would you choose Chef again? I'd say that the exact config management solution you choose is irrelevant as long as you actually choose one of them. What I can say for certain is that we'd definitely still be using configuration management of some sort.
Bearing in mind the existing bias of the fact that we already use and know Chef, it's hard to say exactly what that solution would look like.
After reading your "Our hardware" blog post, we (my co-worker and I) noticed that you don't virtualize much. Why? Generally, we use most of the power of a given machine, so it doesn't necessarily make sense. We have nothing against virtualisation, given the right workload: For example, we give every engineer a virtual machine (since it's unlikely all of them are using the full power of a machine), and continuous integration test nodes (because it's easier to have separate LXC containers to share resources than try and run multiple MySQL/Apache/etc on one host)
That's a very interesting answer. In addition to utilizing the physical box in its entirety, I'm willing to bet you have enough redundancy in your important physical machines to not need some of the fault tolerance features provided by virtualization across a cloud (think Amazon). Excellent point. I think at this point we're 97.3% of the way there having no single box being the only one doing that thing.
No matter how well you make a piece of hardware, it's always going to fail in some way.
For example, we don't even bother putting multiple disks (SSDs) in our webservers, because if one fails, never mind. The DC guys will swap it out at some point and when it's done, 10 minutes later it's back reinstalled and serving the site again.
The remaining 2.7% isn't responsible for production either.
How do you people monitor the Etsy's website from client's perspective, whether you people use some external monitoring services ? We use a number of tools to do this. We use external services, but we also use Lognormal to collect performance data, much like many other sites do.
What RDBMS you people use and how you scale them, ensure high availablity and prevent SPOFs. Mostly MySQL. Check this out: Link to . We also have a little PostgreSQL, but most of the services that use it have been migrated to MySQL with a master-master setup.
What log and data backup solution you people use ? We log a log of data and centralise it with syslog-ng. From that central point we then do a lot of cool stuff with it, like parsing with Logster, sending to splunk, etc.
What are the desired skills required for getting job as Operations Engg. at Etsy? Strong web operations knowledge (linux, apache, mysql, php), configuration management, and networking are all a good starting point.
Who decides the architecture of the application - Developer or DevOps? If both, how do you reconcile the differences? Both. We get together in a room (or over video conferencing) and talk about what would be best for the product, operations, and long term maintainability.
What software and/or techniques do you use for backing up MySQL databases? We use Percona's Xtrabackup, which creates binary backups very fast. We backup all our databases every night to a local store on each Mysql server and then ship those backups to centralized backup servers and to an offsite location.
waves to all The entire Ops organisation is pretty large, and covers infrastructure, security, corp IT, and others. There are 14 people who deal with the production site and network directly. We have about 15 people in the room right now, from a variety of groups who want to say hello!
How big is your team, who all is here today, and how do you manage all the different personalities you work with? Personalities are always great to work with. We enjoy all the different ways of thinking. Sure, sometimes we run into bumps just like any group does. Over time we've built strong relationships with each other and they really help us get over things quickly.
Also, what's for dinner? :P. Dinner is hand pulled noodles :-)
+1 in the Data Center! Keep the noodles away from the servers, dude! Last time I got too close with noodles, the site was flooded with pictures of ramen!
How do you handle routing around failure? Our load balancers do handle the host checking. We run load balancers in pairs, so a hot standby is always ready to go. We don't use DNS internally for any failoveHA stuff. Part 2 of that blog post has yet to come...
Presumably server failures are handled by a load balancer, but how are load balancer failures handled? Our site is fronted by CDNs (three, specifically), so we closely monitor failures they have reaching us. If any of the 3 have an issue, they can be shut off and the bulk of the traffic is moved within a few minutes, thanks to, as you say, very low TTL DNS records. This has some downsides, because of the caching issues/people disregarding any TTL less than X, but those types of CDN failures are fairly short lived.
Do you remove low-TTL entries from DNS? We monitor failures from the CDNs by having them serve a special "Whoopsie" page when they can't reach us. In this we have a "pixel tracker" 1x1 image, which comes back to origin bypassing the CDN so we can track if it could be our issue, or a particular CDN.
Remove anycast entries from BGP? In terms of load balancer failures themselves, we run active-failover and the failover is extremely efficient there. They very rarely have an issue that means our origin is unavailable due to the load balancers themselves.
More information in our upcoming talk at Velocity EU: Link to
What do you guys use to make your MySQL deployment highly available? (master-slave, master-master? where do you write, how do you flip server, roles, etc) We use Master-Master pairs. You can read more details about it here: Link to and here: Link to
What made you go with a semi-home grown backup server instead of an enterprise class storage/backup solution? I know it's probably a bit cheaper to roll your own but I would really not want to test your backups ever. I hope you offsite your backups, please tell me you offsite them. One of the things we love here, is the ability to keep things as simple as possible. That includes our software, our processes and our changes. We based our solution on that. It started off as a collection of shell scripts, and eventually was rewritten in Ruby. It's still incredibly simple, and easy to debug when things break, and affords us the flexibility we want. As an example, we can add very custom modules to test backups to ensure they're correct.
Do you guys version control your schema changes? It seems to be one of the few areas people often don't version control. We don't VC our schema changes. We do have a lot of tests that run against the production schema when code is about to be deployed.
Do you use Redis in production and, if so, how do you achieve high-availability with it? We do use Redis in production, but currently on a limited basis (we have a number of experiments that are either underway or being planned). We maintain a masteslave pair for availability and will be taking a closer look at Sentinel to determine if it's a fit for us.
What do you run your own servers instead of using cloud IaaS providers? Did you do a cost comparison? Is there a non-cost reason? Back when Etsy first started, the cloud was very much in its infancy. At that time everyone was running things on their own platform.
Since then, we've invested heavily in the skills and infrastructure to keep growing our own platform. Certainly, I don't think there's a strong technical reason we couldn't run on a cloud IaaS platform, it's just that we're not there. The setup we have now runs really well and there isn't a strong technical benefit to moving.
We do have a lot of internal virtual machines which developers use, but most of the rest of our hardware is dedicated to specific functions.
How many "fires" do you go through a week or month? What has been the most interesting/stressful situation your team has been through keeping everything up and running? We average about 70 alerts per week at the moment. Those range in severity, of course, but there is the occasional site outage. We work really hard to generate visibility into how our infrastructure is operating; sometimes, knowing which set of dashboards and/or metrics to review in a given situation makes responding to outages a little stressful, but in the end, that level of visibility helps us to respond very quickly when outages occur.
Along these same lines, how do you make sure you have a high signal-to-noise ratio? In other words, how do you prevent the typical "oh, that's just the load balancer acting up, you can ignore the alert"? This is an excellent question. Alert fatigue can be a big problem. The key, once again, is to constantly review and tune the alerts; perhaps a threshold needs to be adjusted; perhaps the alert doesn't need to page (but an email would still be warranted); perhaps the original assumptions that went into designing the alert no longer hold and the alert can be removed. Remember to discuss your ideas about modifying alerts with your team; they'll provide valuable feedback.
I imagine you don't want to be alerted of everything, but also don't want to reduce the number of alerts to a point where you'll miss something. Continuous care and feeding of your alerts is critical because you definitely don't want engineers to be in the habit of ignoring alerts. Otherwise, why have an alerting system in the first place?
To the DBAs, how do you handle high i/o? Also, how long do you test before pushing to prod? Most of our data is sharded, so we divide the IO load among many servers. When we need to expand our Mysql infrastructure, we add more master-master pairs and have internal tooling to move data between shards. Regarding how long we test before pushing to prod, it varies a lot. We have the ability to "ramp-up" new features to a percentage of users. That way we can see how the new feature is impacting performance.
I didn't know MySQL had a Master-Master config. Thought that was an oracle and sql 2012 thing. Mysql doesn't have any restrictions for doing it. You just setup bidirectional replication among two servers and it works. You do have to be careful with auto-increment fields, which we don't use at all.
Mysql doesn't have any restrictions for doing it. You just setup bidirectional replication among two servers and it works. You do have to be careful with auto-increment fields, which we don't use at all. How do you shard the mysql data ? What tool you use to achieve this ? Whether the sharding is on table level or on row level ? How do you handle in case one of the master goes down from a MM cluster ? We shard based on users and shops. All data for a particular user or shop will be all in one shard. When a server goes down, we remove it from our application configuration file. From that point on, all web servers don't know the downed box exists. After that, we bring up a spare box, restore it from last night's backup, let replication catch-up and put it back into the configuration so it starts taking traffic again. While the box was down, the other side of the MM pair took the load of both sides. You can see some additional information about our sharding architecture here: Link to
Good to know. I'd be interested to see some i/o numbers from the DB farm, but I get why you couldn't post those. Thanks for the reply. We currently have around 100 Mysql servers, all of them are 16 disk 15k RPM setup in RAID 10. Each one is capable of around 2000 iops.
What is your server naming scheme? (if you can disclose that without security implications) We name our servers after their functional purpose. Rather than including the location of the server in the first part of the hostname, we use a fully qualified internal hostname to indicate the location of the server.
So basically,
Is it easy for you to tell what is on the server for alerts? When goes down do you actually know why? I ask because it is a problem I am dealing with currently at work. Also how descriptive are your functions and do you have random app38 servers? We logically group our server functions in Nagios and Chef by functional group, ie all of our webservers will be under a webservers host group in Nagios, and our alerts are configured accordingly. Similarly, all of our webservers use the Chef role "Web". We don't usually have the problem that we have a random app38 server and aren't sure whats on it - if you wanted to establish what was on app38, you could go to Chef and look at the "App" role.
The kind of continuous deployment that companies like yours are doing is, I think, still out of reach for a lot of companies, especially ones that don't have big enough in-house devops teams to produce the systems. What tools or services do you think are still not available that would help companies get there more quickly? I think you're right. I spoke about this at length at LISA '11 and the same question came up there.
A lot of CD is to do with culture, much more than tools. I know people who do it with Deployinator, Jenkins, Dreadnot, and a bunch of other things. But the culture is really what makes it. Once you have the culture moving in the right direction, where developers are happy pushing code and owning software problems, and operations teams are OK letting go of the control and working with developers, the tools become less important.
(I realise this isn't directly answering your question, so I'll summarise with: better CD tools like Deployinator and Dreadnot, and betteeasier unit testing things are always good.)
How do you handle automation and change control on the very uncooperative lower layer stuff like bios and raid card config? BIOS: We've managed to avoid it. Which is irritating because recently we discovered that having Hyperthreading enabled was actually hurting performance and scaling, so we had to manually restart tens of servers to fix them, but that was a one off so it was quicker to do it by hand than automate (I'd love to hear if there was a good way to do this if anyone has any ideas...)
RAID cards: We have a set of tools that allow us to burn-in, configure and then install machines automatically. Part of this includes a tool which PXE boots a CentOS "live CD", so we can have full access to the various RAID configuration tools. You can essentially choose a RAID configuration on the command line, and the server is rebooted and reconfigured by loading a configuration file from the master server. No more touching MegaCLI or hpacucli :) More information about how we do this available here: Link to
With HP servers you can use the SmartStart scripting tool and a web hosted XML file to configure the BIOS. Specifically the HPRCU command (formally CONREP). Link to If you'd like to hire me I'd be glad to set it up for you. :) FYI you can also configure iLO with this tool. Thanks! I was aware of the HP methods, but in this case it was Supermicro.
What do you use for an op's dashboard? We also use Nagdash which was written in house (cough) for a Nagios dashboard both in the office and in browsers
What are key operational metrics? How do you deduce business KPI's from your data and how do you instrument them? As for key metrics, we have specific metrics per service (How many DB queries are happening? How about web requests?), and also business metrics.
What hadoop distribution do you run? We're running Cloudera's CDH open source release (specially CDH4.1 right now) Because we use Chef so heavily we decided to stick with the open source release since it's so easy to build new machines, and just install the RPMs/configs.
Have you ever been in a position where PaaS/ outsourced IT was a realistic position for some products/web services within Etsy? How did you manage supporting those products and services? What points of differentiation did you explore for Etsy operations services vs AWS services like Elastic Beanstalk or Heroku? We've certainly considered it. In some cases it has made sense, and in other cases not. That sounds like a cop-out, but we really do take everything on a case-by-case basis. If we were a lot smaller, I would advocate more for out sourcing to help stretch our resources farther. How does it change our options for the future? These are really just some of the questions. With the understanding that our infrastructure is a known good design pattern for us, we try to stick to it where possible.
How much exposure to ITIL have you had? Do you think it's a good framework for new or ever-changing business? (think start ups pivoting, or rapidly bring in new product ideas) How much more (or less) time would it take to support? How much would it cost, vs doing it ourselves?
How do you handle new software releases? Is it a continuous deployment system of some sort? Do you have scripts that automatically push the new to the servers? This is a topic I love to talk about! Due to time constraints I'll will simply say that we have tons of information on this process on our engineering blog at Link to :-)
I've seen a number of presentations that mention Etsy's ability to push to production many times a day. How many man hours would you estimate were required to build that process including the code to do the push, dashboard to track, changes to the code base, etc ? It's extremely hard to say. It definitely didn't happen overnight, we continuously deploy the continuous deployment system :) Many of the things we realised we needed were as a result of failures in the system, and having postmortems to learn what went wrong, and fix those to avoid it again.
I've been looking at the Kale stack a bit recently. Thanks for open sourcing! Do you trust Skyline enough that you have people get woken up in the middle of the night by anomalies it detects? If not, do you expect that you will get there, or is the idea more that it should be a system for showing you which graphs to look at? That's actually what we're working on now - we don't currently generate paging alerts from Skyline, but we definitely want to get there.
What are your favorite types of cookies? MIT magic cookies. No, not those types of magic cookies.
Oatmeal and Raisin.
Expiring cookies?
Triple chocolate.
Would you rather fight one Allspaw-sized Gene Kim, or 100 Gene Kim-sized Allspaws? Depends - which one of them will buy me pizza afterwards?
I said "fight", not "date" ;) Damn! This is tough because I'd rather date them than fight them.
Ok fine: I'd rather take one Allspaw-sized Gene. No-one needs to fight 100 Allspaw. I can barely take on 1 (although I try regularly, I've yet to find his weak spot - next time I'll try after he's had a big lunch, it might slow his thought process down just enough).
Dating co-workers is bad, bad! If that were the case, I would never have met my wife!
(full disclosure, brandyvig is my wife ;-) )
From another comment here I assume you are using CentOS. How do you handle yum upgrades across your servers? The answer is "very carefully". We keep software versions pinned to one version for the most part. When we need to upgrade it (new features, bug fixes, etc), we test it one a small number of systems and then let chef upgrade it everywhere.
also, do you use Perconas MySQL fork? If so what do you think of it compared to standard MySQL? We opened source our chef-whitelist library which we use to accomplish this.
I remember reading on Code as Craft a while back that you were leveraging schooner memcache/membrain for cache replication in the Etsy stack. Is this still the case or have you switched to a less "black-box" solution since? Actually, we've never used the Schooner Membrain at Etsy. We only use Memcached and the caching logic is built into our in-house ORM.
I heard you guys were looking at MongoDB for some stuff. I've experimented with MongoDB at a very large site and it was a dismal failure. If you have looked into it what's your opinion? We experimented with MongoDB for a large project a few years ago. We later decided to move the project from there to MySQL. There were a number of reasons for doing this, which related to stability, domain expertise and wanting to keep our infrastructure more homogenous. That doesn't mean we never add new technologies though. We recently started using Redis because it has some very specific strengths which outweigh the cost of adding another technology to the stack.
Are you guys considering different hardware for your servers? Why not implement HP blade systems or moonshot to increase your density? Have you considered open compute like facebook and other large scale web operations? Opencompute is really interesting. I think we're almost getting to the point now where it might make more sense to go that route vs existing big box retailers, but for now we're in a mostly good place. re: blades, a lot of our HP machines are purchased for their high IO density. if we don't need IO density, we buy Supermicro sort-of-blades, 4 machines in a chassis. More info on those here:
Do you have any hardware pr0n? Link to We do. I'm afraid we can't post any pictures. But matt can explain the beauty :)
Actually we do! Check out our blog post.
What kind of tools have you guys developed for internal use? We have multiple tools, a few can be found in our github page. Some prominent tools that come to mind are our provisioning tools which nicely automates our hosts building process (Everything from configuring RAID/storage profiles to configuring the network interfaces and installing the OS and configuring all services). We also an awesome tool to keep track of each day-today activities, which shows all Jira tickets, github commits for each of the member in our Operations team.
How deeply can you analyze information from sellers and customers? Could you figure out trends from random search terms like "cardigan", "coolstorybro", "u jelly", or "impossibrew"? We have a Data team just dedicated for this sort of things. We use Hadoop.
What are some useful tips / tricks / software I should look into to expand my knowledge? Read stackoverflow and serverfault. Look at what questions people ask about those larger scales. See if you can re-produce the problem and the fix at a smaller scale (maybe in a VM on your computer?)
What software and/or techniques do you use for backing up Hadoop data (to protect it from being destroyed because of a human error)? "Human error" is a term we stay away from wherever possible. Frankly, it doesn't exist as a legitimate reason for problems in complex systems. Yes humans make errors, but those aren't themselves the reasons for something breaking.
We have an internal book club where we're currently reading "The Field Guide To Understanding Human Error" by Sidney Dekker. It explains how you should look at the failure of complex systems. There is something Dekker calls the "Old View", which includes things like human error as a reason for a problem. The "New View" takes the time to look into the actual cause of a problem. For example, let's say someone logs into a production server, thinking it's a development server, and wipes the disk.
The problem isn't that the person did it, but that they were able to do it. Why did the system not make it more clear that it was a production machine? What other safe guards should have been in place to prevent this? Did they need some kind of confirmation before doing it? Why did they need to take this particular action in the first place? And so on :-)
I highly recommend the book, and John Allspaw's talks on postmortems (part 1 of his 90 minute talk from Velocity 2011 is available here)
Why is your security team so cool? It's all the animated cat gifs.
Hey Guys, I love the way you operate and the community seems awesome. I'm currently a sysadmin with programming experience and want to get into devops. Would you guys be willing to hire on someone who's quick to learn and really interested in what you're doing? :) We hire people with a wide range of backgrounds, and look for the best fits for each position :-)
I would say that you can't really get in to "devops" as a job. That would be akin to saying you want to get into "teamwork". It's just a thing you do :-) But having both programming and operations skills is a very good combination and one that is quite high demand these days!
Backing up and restoring any DB in a Continuous Delivery environment is extremely challenging, I haven't heard of anyone that has implemented a solid solution. It would be great if you (Avleen) can touch on this topic a little bit also. We are able to backup and restore live Mysql servers because all our Mysql servers are Master-Master pairs. This allows us to take down a server and keep running. More detail here: Link to
Last updated: 2013-08-16 13:16 UTC
This post was generated by a robot! Send all complaints to epsy.
submitted by tabledresser to tabled [link] [comments]

Why Is Digital Options Better Than Binary Options? - YouTube Forex Vs Binary Options in 2018: What is more profitable ... Difference Between Forex and Binary Options Trading ... Adept's Vault Replication with Binary Differencing The difference between Forex and Binary Options Trading Binary Options Fraud: Inside A Multi-Billion Dollar Global ... Binary fission and mitosis

Make sure you enable the option called – Enable binary differential replication. Click the next button; NOTE! – Select NO Deployment Option if you want to use Microsoft cloud instead of on-prem DPs for patching. Create ConfigMgr Automatic Deployment Rule ADR SCCM Distribute the Package. Select the Distribution Point you want to distribute the package. Click Add and Select the available ... BDR was chosen only because I was unsure what options would be the most useful. If there are options you are using that you would like to see in this tab, shoot me a message. I am open to suggestions. Is a bitmap of the schema generation option for the given article. schema_option is binary(8), and can be the (Bitwise OR) product of one or more of these values. When publishing a table or view, replication allows you to control the object creation options that are replicated for the published object. Schema options can be set programmatically when an article is created using replication ... Compare binary replication tools for PostgreSQL for features and ease of use. This document should classify and differentiate binary replication tools for easier selection and fit to purpose. Comparison Matrix. Tool Documentation License Makes base backups Makes base backups from replicas Manages backups Creates replicas Monitors replication delay Supports automated failover Transport used ... On the Data Source tab: Enable the “Enable binary differential replication” and “Deploy this boot image from the PXE-enabled distribution point” options; On the Optional Components tab: Add the required additional components (In this case I added the PowerShell component) Save the adjusted properties; Configuration Manager will ask you to update the distribution points: Choose Yes ... Binary Differential Item Functioning - Getting Started; DIF - Background. In any discussion of differential item functioning, it is important to note that DIF is not synonymous with bias or fairness. Questions of fairness in tests and measures are directly related to the use of an individual’s score on that test or measure. For example, if men and women from a given nation receive different ... Since a binary option is similar to a call spread, a binary call spread offers a risk reward similar to a condor Sep 01, 2014 · The replication mismatch 15 of the binary call option replication DEK method motivates me to construct an improved static hedging strategy with plain vanilla options and European binary options for hedging an up-and out call option. Jul 20, 2020 · Jun 06, 2019 · An ... Details gap options: the deel on option replication differential binary a gender deposit depends on the interval-level trades of a little pair, but is technically affected by a plugin asset of chapter investments, which may be simple or close. Het trade nature neemt slecht credit sum charge inequalities problematic gemeenschappelijke weather trader nature. Because daily, following a economic ... Hello! As i understand, when "Binary Differential Replication" is checked, then when you make changes in a package and then updates the distribution points, only the changes will be transfered and therefore you wont use so much bandwidth as would have to if you didnt have this option checked. Replication in Drizzle is very simple and multi-source replication is supported. For a walk through of multi-master (multi-source) replication see David Shrewsbury's excellent post here.Because it was very succinctly here I am quoting a lot of his provisioning a new slave post on replication here.But I have added in some detail on the slave.cfg file for clarity for newbies like me, as well as ...

[index] [8301] [27274] [20438] [13360] [26194] [13148] [8732] [3152] [15982] [15479]

Why Is Digital Options Better Than Binary Options? - YouTube

In this video I talk about a question I get alot and that is what is the difference between forex and binary options? In the video I go through the mains difference which are below. Recommended Binary Broker: (Multi-Regulated, Variety of Options, Quick Withdrawals) Note: Broker currently NOT accepting EU ... The Difference Between Forex and Binary Options Trading - Binary Option vs Forex Trading Tutorial. Follow the link below to create a FREE Practice Account: h... Discussing types of copying cell division (binary fission and mitosis and why they are different). Also discuss chromatin vs. chromosome DNA forms, then go t... Forex Vs Binary Options in 2018: What is more profitable? 🔴Want to learn how to trade properly go here: In this video we're going to lo... Adept Vault Replication allows multiple geographic locations to share a common Adept Vault, giving users at any location fast, local access to files. With our Binary Differencing option, only the ... This session will look at binary options fraud, which has become a global phenomenon, particularly in Israel, where many of the perpetrators are based. We wi...