SQL dbo.Chris

Say you wanted to grow the transaction log of a database from a few hundred megabytes to 100GB or even 500GB on 2 striped 1.73TB NVME drives in a 4 disk pool using Storage Spaces direct.  First this is something you would NOT normally do, but if you had some some time on a new fast server these are the sort of fun crazy things you do in your spare time.  Simple math shows that the Local NVME storage was 2.2 Times faster than the SSD SAN based storage connected over 40Gb FCoE. In this example: I grew the log file 500GB on an I3 AWS Local NVME instance that took 8 Minutes and 32 Seconds In this example: I grew the log file 100GB on a Dell Compellent SC8000 with an SSD backed lun that has around 15k Write and 30K Read IOPS taking around 3 minutes and 55 Seconds

I had some time at AWS re:invent to play with these I3 instances and a few others which are designed for I/O intensive workloads and are equipped with super fast NVMe SSD storage.  These instances can deliver lots of fast read and write IOPS making them a great fit for any workload that requires high throughput and low latency including relational databases for Microsoft SQL and Oracle databases server. When compared to the I2 instances, I3 instances deliver storage that is less expensive and more dense, with the ability to deliver substantially more IOPS and more network bandwidth per CPU core.  Additional, if you take advanatage of the  I will show some of the Read and Write benchmarks using the  Diskspd Utility : A Robust Storage Testing Tool (superseding SQLIO) with DiskSpd Batch that was created by HeraFlux Technologies which can be found HERE    as well as the quick button push of CrystalDiskMark.  So if you want to see all the numbers ...