Reference Architecture for Content Management on Azure HDInsight with HBase

TSG has recently been working with a multiple clients that are deciding on whether to deploy their Content Management platform in the Azure cloud from Microsoft or on premise. The discussion typically boils down to “how much would it cost for me to deploy this in the cloud vs procuring hardware and software licenses for an on premise deployment?” This post will provide a reference architecture and sizing estimate to help understand the various components and their relative costs to procure, support, and maintain on Azure.

For many of our clients that are struggling with the idea of deploying some of their content management applications in the cloud, the conversations can hit a speed bump when attempting to estimate how much a solution will cost to deploy and maintain in the cloud. Microsoft, AWS, and Google all offer calculators to estimate monthly costs, but depending on who is filling them out, there can be radically different results based on the assumptions for how the solution should be deployed.

For clients that are evaluating using TSG’s OpenContent Management Suite on HBase, we have provided a reference architecture diagram below for the individual pieces of the solution and where they map to the Azure offerings.

It can also be unfair and difficult to make an apples to apples comparison on price alone when making decisions to deploy on-premise vs on a cloud provider. Just because it only costs you $2000 once to buy an on-prem server doesn’t mean that is the only cost for an on-premise deployment. Other things that should be considered when estimating the overall cost of ownership include:

• Ongoing support costs – This is often the largest “hidden” cost of an on-premise deployment as the staff and physical space to maintain the infrastructure aren’t free. Various ways for estimating this cost include a chargeback model or a complicated formula based on headcount to support based on the number of servers/storage size, etc. For cloud providers, this is built into the cost of the service.
• Backup/Recovery – Most cloud providers offer 99.99% uptime in their SLAs, and for the .01% of the time that there are issues, there are built in fail-overs baked into the cloud architecture and the costing that don’t require internal resources to come up with and test your DR plans
• Speed of Deployment – We have worked with some clients that have said there is a 6-8 week time frame for completing new on-premise infrastructure requests. We can often perform a complete migration and rollout of a new platform in that amount of time, so the barriers to entry are often much smaller to quickly prove out new technologies.
• Flexibility – If the above 6-8 week lead time for servers was sized incorrectly, it could be another 6-8 weeks to change or add to the order. With cloud providers, the changes are instant to scale up or scale down as needed.
• Cloud service offerings – Azure and other cloud providers often provide services like an HDInsight HBase cluster that allow for simple hosted and supported solutions. There are zero additional costs associated with the patching/upgrading/maintenance traditionally provided by on-premise support staff needed to support these servers since they are included in the costs of the cloud service. Similar offerings for traditional RDBMS like SQL Server are also offered by cloud providers that can often eliminate the need for dedicated DBA teams to maintain the servers and indexes.

Please click below to download the costing estimate or to contact us for more information: