Cloud Data Migration Performance Problem At Leading UK University
Regional Support Director, Global Equipment Manufacturer
A leading UK university was suffering severe performance issues when migrating data to cloud storage. Poor performance was hampering data replication and impacting critical systems throughout the university infrastructure. For more than a year technical support teams had struggled to find the cause of the issue and plans to expand online learning programs had to be put on hold.
The problem was constant and therefore very easy to reproduce in a controlled environment. This meant that we could move quickly to establish a diagnostic plan.
Using our standard structured approach, we devised a Diagnostic Capture Plan to capture network data at multiple points along the cloud data migration flow. We captured data showing the migration process from start to finish. This enabled us to create a clear picture of exactly what was going on during the movement of data between the university systems.
Analysis of the data showed that the problem was caused by delays in generating TCP acknowledgements from the cloud content platform. The university infrastructure was performing well but the cumulative cost of the TCP acknowledgement delays meant that data uploads were taking up to 20 times longer than they should have. On average, more than 95% of the upload time was attributed directly to the TCP acknowledgement delays.
We passed our findings and evidence to the technology vendor and worked with them to identify the specific equipment causing the problem. They agreed with our findings and dispatched support engineers to the site to test and repair the defective equipment. The university data migration performance issue was resolved; solving a yearlong technical problem that imposed significant resourcing and financial costs.