This post is more than 5 years old
4 Posts
0
1182
Is there a way to use SRM workbench for multi-region config
Can anyone tell me how to use the SRM Designer workbench to size a multi-region setup?...ie Main Vapp sits in one datacenter but will have collector/s in many data centers. The designer workbench is a great tool but seems to work on the assumption that all collectors and all devices are essentinally in the same spot so just loading all the cumulative device metrics will not work. While I would guess I could run it for every region using the worksheets to have it size out collectors, that would not only be cumbersome but i would not then take into effect the rollup needed for potential required additional backend expansion.
Also, can someone tell me if there is a 4.X version of designer workbench available to customers and/or partners as the only version publicly available appears to be 3.7
many thanks
john J
johnjaworski
4 Posts
0
April 11th, 2017 16:00
Awesome tips Kens
Many thanks
Sent from my iPhone
fredde1
120 Posts
0
April 10th, 2017 10:00
Hey Johnjaworski:
Can you please clarify your statement:
The designer workbench is a great tool but seems to work on the assumption that all collectors and all devices are essentially in the same spot so just loading all the cumulative device metrics will not work
The device metrics should not be contingent on the data center that the collectors are at. If your collectors can all communicate with the vapp/infrastructure, then it should be fine, within the milliseconds allowed( no network latency)
Thanks
Tim
johnjaworski
4 Posts
1
April 10th, 2017 17:00
Thanks for your input.
It is absolutely site dependent from what I can tell. Example. If I have 6 different regions and 3 of those data centers are larger with 10 or more 10 Vmax arrays , 10 VNX arrays, Vplex. Etc in half of the data centers and half of the data centers are smaller and I plug that cumulative information , in mass in the workbench metric spreadsheet, it will output X number of collectors for those metrics. The tool assumes that load is all in one data center and essentially “levels it” when it creates collectors without regard to where that collector resides, when in fact, the more logical design (and the one we must follow due to networking restrictions) is that each DC has a local collector that pushes back to the main VAPP. In real life a larger data center might very well have own dedicated VMAX collector for example due to the number of regional Vmax devices whereas a smaller DC might have a single collector for Vmax/vnx/brocade..etc
In that case, as I mentioned, I could run the workbench tool once for each DC but that doesn’t really fully represent the configuration either given that the FE/DB/BE are in the primary DC with regional collectors so running it separately will not create the additional BE if the total metrics from all the collectors deem it necessary.
What am I missing here?
John Jaworski
Senior EMC Storage Engineer: Midrange Storage
ATOS
John.Jaworski@atos.net
UK_SA
9 Posts
1
April 11th, 2017 07:00
Hi John.
I encounter this issue frequently and there is a straightforward way around it.
First, make sure it is in automatic mode and add all the devices you want to discover, but make sure you add each site quantities separately. For example, if you have 2 Cisco switches in site A and 2 in site B, then add 2 lots of Cisco switches - 2 switches in each. You'll get 2 Cisco switch collectors on screen. Repeat for all site devices, separating as required.
Then, switch the design to manual mode and start moving the collectors around so that you consolidate a certain site's devices on 1 or more Collecting VMs - which become that site's local Collecting requirements. You may have to temporarily increase some of the Collecting VMs memory, to give you head room to move things around - remember to reset after you have finished. You may have to add additional Collecting VMs too, but don't do that until you really have to, as they can't easily be deleted.
When done, you should be able to have everything in one SRM infrastructure, but easily identify separate site Collecting VMs. On some of the Collecting VM graphics, you can even edit the names to make more sense.
Hope this helps.