October 2020 Sprints – Sizer Wiki Production

Oct 19

hi everyone

We went live last night with latest. Both big and small changesOn small but good things

Updated Oracle sizing to match the recent changes in SQL Server production cluster sizing. We already had dedicated cluster as requirement fr Oracle but now 1:1 CVM with total of 12 physical cores (yep we want lot of I/O capability) and min of 14 cores and 52 specints.
Align the VCPU:pcore ratio when doing either configuration or performance sizing with Collector
Bulk edit can now be done for XenApp or SQL Server workloads
Robo model addition: NX3060-G7
DX Mine appliance: 1.92TB SSDs – RI

On BIG things

Sizer FINALLY has I/O !! Well technically we had it for Files Application sizing but not general purpose use. Now have I/O performance sizing for both Cluster sizing (Raw) and Server Virtualization workloads. Where historically we would size for capacity, now can size for I/O and Capacity (whichever is greater requirement)

Want to thank both @ratan.kumar and @scott.fadden for all their hard work in getting the I/O effort going. There was a lot of testing and analysis to get this scoped. They both worked very hard and is excellent work. This is what I love to see in Sizer so it is a better for you all

Here is the I/O panel in the workloads

Oct 6

Hi everyone, we went live with the current sprint, below the major highlights:SQL sizing enhancements: Major changes to this one..

Changes to the Business Critical ON/OFF options, default settings. Default SQL sizing to be business critical.
Sizer to allocate additional CVM cores(1:1 vCPU to pCore ratio) to aid in performance for business critical option.
Business critical SQL workload to be In a dedicated cluster, with only other SQL workloads. VDI, Server Virt etc not allowed to the SQL dedicated cluster.
All Flash or NVMe models only with high frequency processors for higher performance

Budgetary quote : HPE-DX

Now, generate budgetary quote for sizings on DX. Earlier budgetary quote would show only SW/CBL quote but now HW BOM price estimates also included
The HW BOM quote covers complete BOM including PSU, transceivers, chassis etc and including HW support prices

Files changes:

New File SKUs with tiered pricing is supported now, including generating Frontline quote through Sizer.Sizer’s budgetary quote for Files is also updated with newer SKUs and pricing approach.
Application storage – updated with latest performance numbers across hybrid and AF nodes.
With increased throughput/IO per node, would need fewer nodes than before for same workload.
Defaults to 1x25GbE NIC for smaller nodes, 2x25GbE for larger nodes.

Collector/RVTools

Now can choose to size for storage based on VMs consumed or provisioned capacity during import.

Usability

Era quoting in Frontline supported through Sizer
Bulk edit – now also supported for Oracle, Backup workloads
HPE-DX default NIC recommendation -FLOM/OCP in both auto and manual
Updates to XC models, updated list of CPUs and SSDs across models

Thanks Ratan. In regards to SQL, it has grown up in Sizer sort of speak. If you are looking for adding a small sql database in a cluster of say server virt, then go with Business Critical off.Then SQL workload can be in a cluster with other type workloads, we take 2:1 CVM and no min cpu in terms of cores and specintsGo to Business Critical and then it is a dedicated cluster, 1:1 CVM with total of 12 physical cores (yep we want lot of IO capability) and min of 14 cores and 52 specints. Config is also AF or NVme. We will be making the same changes for Oracle in current sprint.

Leave a Reply Cancel reply