Microsoft Azure Synapse Analytics combines data warehousing with analytics, machine learning and visualization. It brings together services which, in the past, were separate, under one umbrella. For example, Azure SQL pools (warehousing) and Apache Spark pools (analytics and machine learning) are bundled and made available as part of a unified service (or perhaps we could call it a suite).
Synapse is a powerful toolkit so it’s not surprising more companies are adopting it for a variety of use-cases. Not all of these use-cases however, are a good fit for Synapse and, considering the potential expense, it’s a good idea to do a lot of homework (good old fashioned due diligence) before running that deployment.
Take a look at this pricing diagram:
The sample scenario shown, which depicts a fairly modest configuration and usage profile, generates an estimated 6,904.10 USD per month in spending. The basic database pool unit for determining the performance tier and, therefore, the cost, is the Data Warehouse Unit:
One of the key decisions to make is what DWU tier is necessary for the use-case; this, like all architectural choices has a direct impact on cost so taking your time and using decision aids, such as this one, is strongly recommended.
Is This the Right Tool for the Job?
Synapse is very compelling for many organizations because it seems to hold the promise of delivering everything needed from a warehousing and analytics point of view together in a nicely wrapped package. No doubt, this is why I’m seeing more clients eagerly building Synapse environments.
There’s a catch however…many of these same clients really only need a database and a Power BI connection for visualization and are only scratching the Synapse service’s surface while paying substantial bills.
Before deploying Synapse, ask a few basic questions:
1.) Do we need analytics and warehousing or just one or the other service?
2.) Are we running up against the limits of our current environment (for example, if it’s on-premises, are we routinely struggling to add more storage and compute to meet the need)?
3.) Is the likely expense matched by the expected return on investment (for example, is analytics at the scale Synapse offers core to the business and part of the value chain in some tangible and measurable way)?
4.) Have we performed a cost benefit analysis to determine if choosing Synapse over other solutions (for example, other Azure services that can be coupled) is the right move?
5.) Have we determined how we’ll include Synapse’s runtime cost into our FinOps practice ?
If, after doing your homework you still think it’s a good idea to invest in Synapse, the next step is to build an architectural (and therefore, cost) profile that suits your purpose and cost tolerances. The Azure Pricing Calculator is a key resource here.