High Availability in Cloud Computing Platforms

By Asenika Liyanage  10/31/2020 

Cloud Computing has been a trending technology since its introduction in the last decade. Cloud ensures transparency of all the Hardware and Software resources, allowing access through an advanced interface. In addition to this, Cloud resources attract users through effective cost management of overheads, resource management, security like aspects as well.  




Availability and Cloud Outages: Cloud Computing Platforms

Due to the highly rising popularity and demand, there are novel aspects to be considered in this arena. Availability of Cloud Resources and controlling Cloud Outages would be one such aspect.

Cloud outages or Service down-times affect the Quality of Service provided as well as the reputation of the Cloud service providers. Thereby, directly causes loss of revenue and customer turn overs. According to IWGCR, Cloud Down Time per hour causes a loss of $336,000. This amount varies from Business to business according to their dependability on Cloud Services in their Core Business processes. 

Therefore, most Cloud Services are establishing different strategies for ensuring High Availability (HA) in their services. Some of the Top cloud service providers such as AWS, Microsoft Azure and Google Cloud are concerned with improving the High Availability in their Cloud platforms. 

Basic Concepts: Ensuring High Availability

There are three key concepts in ensuring High Availability in any system, as, 

  • Redundancy 
  • Monitoring and 
  • Fail Over
Redundancy is having additional components of mission critical systems/components standby for the purpose of taking over in case of a system/sub system failure. In order to do this taking over, current system should monitor all its processes to detect any anomalies where a component may cease to work as expected. With correct monitoring, any failures can be detected. Thereafter, an automatic fail-over system should be setup for the redundant components to take over the work in progress.  

Apart from these concepts, some technical factors are to be weighed in as well, to ensure smooth fail over and continuation of services with minimal interruptions and highest possible transparency at the users' end. 

Proper and timely data backup strategies and recovery mechanisms must be implemented.  An automatic systems of data backup to a secondary location, which can be recovered back has to be used during the fail-over process, to ensure error free, redundant system.

A Load Balancer is used with the aim of managing traffic by routing it through systems, when a system has failed/not responding. It would be aware of failures and automatically route the traffic to available systems., through monitoring.   

Finally, Clustering of similar nodes is done to ensure Transparency of different distributed nodes to users. The entire cluster is considered as one unit, where one node can fail-over to the next node in case of failures.  

This article is an overview of the key concepts related to ensuring High Availability in Cloud Platforms as well as similar Distributed Systems. For further reading on  implementations of these concepts refer the below links. 


Further Reading and References 



Comments

  1. Suranga WijayarathnaNovember 22, 2020 at 9:06 PM

    Good writing Asenika. Learned a lot about High availability in cloud platforms.
    Could you please explain me more about any limitations in cloud platform servicess in facilitating high availability?

    ReplyDelete
    Replies
    1. Asenika LiyanageNovember 27, 2020 at 5:04 PM

      Thank you Suranga. Yes, there are certain limitations specific to these platforms, when implementing HA. These limitations are related to security, privacy and also to the service availability. You can read about such limitations and potential solutions as well in the following research.

      https://www.researchgate.net/publication/305469666_Cloud_Computing_A_Survey_on_its_limitations_and_Potential_Solutions

      Delete
  2. Ruvy RathnayakeNovember 28, 2020 at 12:18 AM

    Though cloud platforms are highly available, the control over these cloud platforms are comparatively lesser than the physical systems. Isn't it?

    ReplyDelete
    Replies
    1. Asenika LiyanageDecember 8, 2020 at 8:15 AM

      Yes Ruvishka. Comparatively lesser control than a normal system. Main reason is because it is of a third party vendor. We get limited access, customization, reliability etc. But, modern platforms are tackling with these issues in different ways.

      Delete
  3. Osura WeerasingheNovember 28, 2020 at 5:59 AM

    Nice article friend. Does clustering means independent group of resource that can work in isolated manner?

    ReplyDelete
  4. Rajitha GamageDecember 11, 2020 at 2:41 AM

    Good read Asenika. According to your review how do you rate cloud service?

    ReplyDelete

Post a Comment