Skip to content

Latest commit

 

History

History
 
 

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 

readme.md

Before you begin

Pre-req

  1. Install dysk kernel module and dyskctl cli
  2. dysk depends on Azure Storage (Page Blobs). Follow the below steps to create azure storage account via the cli

Install and configure azure cli

  1. [Install azure cli] (https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest).
  2. Login with az login
  3. Set your active subscription with az subscription set -s {AZURE-SUB-ID}
  4. Check your active subscription with az account listi -o table

Create storage account

Names etc..:

export RG="mydyskrg"
export ACCOUNT="dysktestaccount01"
  1. Create a resource group with az group create --name ${RG} --location westus
  2. Create a storage account with az storage account create -n ${ACCOUNT} -g ${RG} -l westus --sku Standard_LRS

If you are running dysk locally then choose azure location closest to you. If you are running on Azure then choose the same azure region where your test VM are.

  1. Get account key via ```export KEY="$(az storage account keys list -g ${RG} -n ${ACCOUNT} | jq -r '.[0].value')"

Premium vs Standard Storage Accounts SKUs

Details here: https://docs.microsoft.com/en-us/azure/azure-subscription-service-limits tl;dr: Premium SKU will give you higher iops for larger page blobs (Highest iops achieved at 4TB page blobs). Standard accounts will give you flat iops for all page blobs sizes but but with lower cieling and capped at the storage account level.

How can make the best selection

  1. If you want fastest iops you will have to go for largest page blob on premium sku.
  2. For any other iops go for standard sku and span them over large number of accounts

max # of accounts per subscription is 200 account (250 with a support request).

Enjoy the Examples!