Evaluating Performance for Cloud Data Warehousing, Part 5- Conclusion

In my last four blog posts, I’ve outlined a good evaluation process for cloud data warehousing, a process that should help you prove out the value of moving your data to the cloud and to get buy-in from your executive sponsor.

Evaluating Performance for Cloud Data Warehousing, Part 5 - Wrap it all up and go!

Let’s quickly review the key steps of that process as a conclusion to this series of blog posts:

  1. Test at a scale that is going to be realistic for your production workload

  2. Your best option is to use your actual workload if you have it available, but if you don’t, then cherry pick from standard benchmarks like TPC

  3. Test individual query performance both on a quiet system as well as as a busy system

  4. Test throughput under concurrency

  5. Factor in the costs… both for platforms that charge per second/minute/hour, or for platforms that charge per query

  6. Evaluate the ease and impact of scaling the cloud platform up or down, or using multiple clusters if that feature is available. Does the platform gracefully handle spikes in workload? Can you minimize the impact on query performance when loading data? Can you save money when the platform is at rest?

TESCHGlobal has helped many organizations evaluate and implement cloud data warehousing as a means of improving performance and decreasing costs. We can provide the thought leadership, technical guidance, and engineering to help your organization make this leap, whether this is your first jump into the cloud or an extension of an existing cloud strategy. Please contact us to learn more!

Want to review this series in depth? View the previous posts:

Part 1 - Design

Part 2 - Execution

Part 3 - Compiled Results

Part 4 - Evaluate Elasticity