Posted By: Tony Baird
Last Updated: Tuesday November 16, 2021
At Hawk Host we’re always looking at opportunities to improve performance of our services while providing minimal to no impact on running services. This can mean performance improvements in software but also additional servers or entirely upgraded hardware. In Singapore we had an opportunity to do a major hardware upgrade.
When our Singapore location was initially deployed it was powered by Dual Intel Xeon E5-2620V4 (Broadwell) and saw some upgrades to Dual Intel Xeon Gold 5218 (Cascade Lake) processors. Our SSD redundant storage was powered by a large number of Samsung SM883’s. All of this hardware was providing a high performance redundant environment to both our Shared and Cloud compute customers alike. We however saw an opportunity to replace all our systems with AMD EPYC based systems which our team worked diligently on over the summer and this work has now been completed.
Our Singapore cloud is now powered entirely by Dual AMD EPYC 7402 (Zen 2 / Rome) running at 3.3ghz (all core turbo boost). We also upgraded the storage component moving to the PCI-E Gen4 Samsung PM9A3 a reduction in I/O latency and the potential for increased IOPS. Our storage system is of course still redundant and protected against drive and server failure. This means individual drive failures are almost unnoticeable and hardware failures result in significant reduction in downtime.
While this post is about our Singapore location we have also been hard at work adding new hardware to many of our other locations. Our Hong Kong location continues to have additional Dual AMD EPYC 7402 systems added as necessary. In North America we’ve been adding additional Dual Intel Xeon Gold based systems in the form of Cascade Lake or Ice Lake depending on availability. Of course just like our Singapore location all with minimal impact to running services and no increase in prices even with improved performance.