Resilliancy

Azure Cosmos DB–Multi Master

October 8, 2018 .NET, .NET Core, .NET Framework, ASP.NET, Azure, Azure CLI, Azure Cosmos DB, CosmosDB, Data Consistancy, Data Integrity, Microsoft, Multi-master, Performance, Reliability, Resilliancy, Scalability, Scale Up No comments

During the Ignite 2018, Microsoft has announced the general availability of Multi-Master feature being introduced to Azure Cosmos DB to provide more control into data redundancy and elastic scalability for your data from different regions with multiple writes and read instances.

What is Multi-Master essentially?

Multi-master is a capability that provided as part of Cosmos DB, that would provide you multiple write regions and provides an option to handle conflict resolution automatically through different options provided by the platform. Most of the major scenarios you would encounter the conflict can be resolved with these simple configurations.

A sample diagram depicting a use case of load balanced web app writing to respective regional master:-

image

With multi-master, Azure Cosmos DB delivers a single digit millisecond write latency at the 99th percentile anywhere in the world, and now offers 99.999 percent write availability (in addition to 99.999 percent read availability) backed by the industry-leading SLAs.

image

Wow! That’s an amazing performance Cosmos DB guarantees to provide so that your mission-critical systems will have zero downtime, if they start using Cosmos DB.

 

How to Enabled Multi-Master support in your Cosmos DB solutions?

Currently multi-master can only be enabled for new Cosmos DB instances using “Enable Multi-Master” option in Azure Portal or through PowerShell or ARM templates or through SDK.

These options are detailed below with necessary examples:

1.) Azure Portal – Enable Multi-region writes and Enable geo-redundancy

image

2.) Azure CLI 
Set the “enable-multiple-write-locations” parameter to “true”

az cosmosdb create \
   –-name "thingx-cosmosdb-dev" \
   --resource-group "consmosify-dev" \
   --default-consistency-level "Session" \
   --enable-automatic-failover "true" \
   --locations "EastUS=0" "WestUS=1" \
   --enable-multiple-write-locations true \

3.) AzureRM PowerShell
In AzureRM PowerShell cmdlet – Set enableMultipleWriteLocations parameter to “true”

$locations = @(@{"locationName"="East US"; "failoverPriority"=0},
             @{"locationName"="West US"; "failoverPriority"=1})

$iprangefilter = ""

$consistencyPolicy = @{"defaultConsistencyLevel"="Session";
                       "maxIntervalInSeconds"= "10";
                       "maxStalenessPrefix"="200"}

$CosmosDBProperties = @{"databaseAccountOfferType"="Standard";
                        "locations"=$locations;
                        "consistencyPolicy"=$consistencyPolicy;
                        "ipRangeFilter"=$iprangefilter;
                        "enableMultipleWriteLocations"="true"}

New-AzureRmResource -ResourceType "Microsoft.DocumentDb/databaseAccounts" `
  -ApiVersion "2015-04-08" `
  -ResourceGroupName "consmosify-dev" `
  -Location "East US" `
  -Name "thingx-cosmosdb-dev" `
  -Properties $CosmosDBProperties

4.) Through CosmosDB SDK
Setting connection policy in DocumentDBClient and set UseMultipleWriteLocations to true.

ConnectionPolicy policy = new ConnectionPolicy
{
   ConnectionMode = ConnectionMode.Direct,
   ConnectionProtocol = Protocol.Tcp,
   UseMultipleWriteLocations = true,
};
policy.PreferredLocations.Add("East US");
policy.PreferredLocations.Add("West US");
policy.PreferredLocations.Add("West Europe");
policy.PreferredLocations.Add("North Europe");
policy.PreferredLocations.Add("Southeast Asia");
policy.PreferredLocations.Add("Japan East");
policy.PreferredLocations.Add("Japan West");

Azure Cosmos DB multi-master configuration is the game changes that really makes it a true global scale database with automatic conflict resolution capabilities for data synchronization and consistancy.

In my later sessions I will write examples to cover how conflict resolutions can be configured and used in realtime scenarios.

Useful Refs:

Azure Cosmos DB – 429 Too Many Requests

October 6, 2018 .NET, Azure, CosmosDB, Document DB, Microsoft, Performance, Reliability, Resilliancy, Scalability, Visual Studio 2017, VisualStudio, VS2017 No comments

Recently while I was doing Performance Testing in one of the APIs interacting with Cosmos DB, I encountered a problem as Azure Cosmos DB API’s started returning Http Code 429.  Http Status Code 429 indicates that too many request been received or request rate is very large. This error would happen when we have concurrent users trying to write or read from same cosmos db collection.

Following diagram covers the architecture of the performance test I am performing:

image

Based on analysis it found out to be the Throttling happening from Azure Cosmos DB, as we make requests that may use more than provisioned Request Units(RU) per second. We were using default Cosmos DB configuration for a fixed collection of 1000 RU’s per second which is sufficient enough for a 500 reads and 100 writes for a 1 kb file. You can refer more about Request Units from Azure Docs.

image

 

 

 

Solution(s):

1. Now first logical step we can do is to get rid off this error by increasing the Throughput for the collection.  I am going to increase to 10000 RU/s maximum allocatable for a Storage Capacity: Fixed.   This should ideally improve the Throughput for 250 or more virtual users hitting.

image

2. Second logical step is to improve the code: Improve the connection parameters in the Document DB SDK –> DocumentDbClient. For this I referred to the Microsoft Docs: Performance tips for Azure Cosmos DB and .NET

Providing optimum values to the following Properties in RetryOption class   to be passed as parameter to Connection Policy.

image

 

In my case I provided a value of 30 to give ultimate results:

new RetryOptions() { MaxRetryAttemptsOnThrottledRequests = 30, MaxRetryWaitTimeInSeconds = 30  }

That should resolve most of the 429 issues when dealing with Cosmos DB SDK

Azure Cosmos DB – Consistency Levels

June 2, 2018 Azure, CosmosDB, Data Consistancy, Data Integrity, Higher Availability, Microsoft, Reliability, Resilliancy, Scalability No comments

CosmosDB is a planet scale multi model, multi-region NoSQL database service provided as part of Azure Platform. Azure Cosmos DB is designed to provide global distribution for every data model you choose while creating Cosmos DB.  It is promised to provide low latency and various well-defined consistency models to ensure data redundancy and high availability.

In this short diagram I will be covering the different consistency models available with Cosmos DB and their benefits:

image

There said depending on your data criticality and needs of faster accessibility, you can choose between any of the above consistency models. I strongly trusts in session consistency, as it ensures a balance b/w both.  But again it is totally depending on your business case and how critical is your system depends on the accuracy of this data.

Hope you enjoyed this short article!.

Further reads: https://docs.microsoft.com/en-us/azure/cosmos-db/consistency-levels

Azure Functions App–Run OnDemand Serverless code – a path way to Serverless Computing

June 18, 2017 App Service, Azure, Azure Functions, CosmosDB, Microsoft, Resilliancy, Scalability, Windows Azure Development, Windowz Azure No comments

Azure Functions is a new cloud solution from Azure that would let you execute small pieces code or “functions” in the cloud.  This means you do not have to worry about the infrastructure or environment to execute your little piece of code to solve any of your business problems.

functions-logo

Functions can make development even more productive, and you can use your development language of choice.

Benefits:

  • Pay only for the time your code runs and trust Azure to scale as needed.
  • Azure Functions lets you develop serveries applications on Microsoft Azure.
  • Supports wide variety of development language choices , such as C#, F#, Node.js, Python or PHP.
  • Bring your own dependencies – you can bring any of your Nuget/NPM dependencies for your functional logic.

What can we do with Azure Functions?

Azure Functions is a very good  solution for processing data, integrating systems, working with the internet-of-things (IoT), and building simple APIs and micro services.

Functions provides templates to help you  get started with some useful scenarios, including the following:

  • BlobTrigger – Process Azure Storage blobs when they are added to containers. You might use this function for image resizing.
  • EventHubTrigger – Respond to events delivered to an Azure Event Hub. Particularly useful in application instrumentation, user experience or workflow processing, and Internet of Things (IoT) scenarios.
  • Generic Webhook – Process webhook HTTP requests from any service that supports webhooks.
  • GitHub Webhook – Respond to events that occur in your GitHub repositories.
  • HTTPTrigger – Trigger the execution of your code by using an HTTP request.
  • QueueTrigger – Respond to messages as they arrive in an Azure Storage queue.
  • ServiceBusQueueTrigger – Connect your code to other Azure services or on-premises services by listening to message queues.
  • ServiceBusTopicTrigger – Connect your code to other Azure services or on-premises services by subscribing to topics.
  • TimerTrigger – Execute cleanup or other batch tasks on a predefined schedule.

Integration Support with other Azure Services:

Following are the services integration supported by Azure Functions app.

  • Azure Cosmos DB
  • Azure Event Hubs
  • Azure Mobile Apps (tables)
  • Azure Notification Hubs
  • Azure Service Bus (queues and topics)
  • Azure Storage (blob, queues, and tables)
  • GitHub (webhooks)
  • On-premises (using Service Bus)
  • Twilio (SMS messages)

Costing:

Azure functions will be charged based on two pricing plans below:

  1. App Service Plan – if you already have an Azure App Service running with Logic, Web, Mobile or Web Job, you can use the same environment for your Azure functions execution without needing to pay for extra resources.  You will be charged based on regular app service rates.
  2. Consumption plan  – with this plan you only need to pay for how long and how many times your functions runs and computational needs/resource usage during that execution time. Consumption plan pricing includes a monthly free grant of 1 million requests and 400,000 GB-s of resource consumption per month.

You can find further pricing related info here

Support and SLA:

  • Free billing and subscription management support
  • Flexible support plans starting at $29/month. Find a plan
  • 99.95% guaranteed up time. Read the SLA

Useful Links:

Scalability – Scale Out/In vs Scale Up/Down (Horizontal Scaling vs Vertical Scaling)

October 1, 2016 Architecture, Azure, Cloud Computing, Cloud Services, Horizontal Scaling, Performance, Reliability, Resilliancy, Scalability, Scale Down, Scale In, Scale Out, Scale Up, Software/System Design, Vertical Scaling, Virtualization No comments

When you work with Cloud Computing or normal Scalable highly available applications you would normally hear two terminologies called Scale Out and Scale Up or often called as Horizontal Scaling and Vertical Scaling.  I thought about covering basics and provide more clarity for developers and IT specialists.

What is Scalability?

Scalability is the capability of a system, network, or process to handle a growing amount of work, or its potential to be enlarged to accommodate that growth. For example, a system is considered scalable if it is capable of increasing its total output under an increased load when resources (typically hardware) are added.

A system whose performance improves after adding hardware, proportionally to the capacity added, is said to be a scalable system.

image

This will be applicable or any system such as :

  1. Commercial websites or Web application who have a larger user group and growing frequently,
  2. or An immediate need to serve a high number of users for some high profile event or campaign.
  3. or A streaming event that would need immediate  processing capabilities to serve streaming to larger set of users across certain region or  globally.
  4. or A immediate work processing or data processing that requires higher compute requirements that usual for a certain job.

Scalability can be measured in various dimensions, such as:

  • Administrative scalability: The ability for an increasing number of organizations or users to easily share a single distributed system.
  • Functional scalability: The ability to enhance the system by adding new functionality at minimal effort.
  • Geographic scalability: The ability to maintain performance, usefulness, or usability regardless of expansion from concentration in a local area to a more distributed geographic pattern.
  • Load scalability: The ability for a distributed system to easily expand and contract its resource pool to accommodate heavier or lighter loads or number of inputs. Alternatively, the ease with which a system or component can be modified, added, or removed, to accommodate changing load.
  • Generation scalability: The ability of a system to scale up by using new generations of components. Thereby, heterogeneous scalability is the ability to use the components from different vendors.

Scale-Out/In / Horizontal Scaling:

To scale horizontally (or scale out/in) means to add more nodes to (or remove nodes from) a system, such as adding a new computer to a distributed software application.

image

Pros:

  • Load is distributed to multiple servers
  • Even if one server goes down, there are servers to handle the requests or load.
  • You can add up more servers or reduce depending on the usage patterns or load.
  • Perfect for highly available web application or batch processing operations.

Cons:

  • You would need additional hardware /servers to support. This would increase increase infrastructure and maintenance costs.
  • You would need to purchase additional licenses for OS or required licensed software’s.

Scale-Up/Down/Vertical Scaling:

To scale vertically (or scale up/down) means to add resources to (or remove resources from) a single node in a system, typically involving the addition of CPUs or memory to a single computer.

image

Pros

  • Possibility to increase CPU/RAM/Storage virtually or physically.
  • Single system can serve all your data/work processing needs with additional hardware upgrade being done.
  • Minimal cost for upgrade

Cons

  • When you are physically or virtually maxed out with limit, you do not have any other options.
  • A crash could cause outages to your business processing jobs.

We discussed in detail about the both approach in Scalability, depending on the need you will have to choose right approach. Nowadays high availability of cloud computing platforms like Amazon AWS/Microsoft Azure etc., you have lots of flexible ways to Scale-Out or Scale-Up on a Cloud environment, which provides you with virtually unlimited resources, provided you are being capable to pay off accordingly.

Hope this information was helpful, please leave your comments accordingly if you find any discrepancies or you have any queries.