ECC/Performance Manager vs SMC/SPA

not sure who told you that you can't do that. I have one server that runs ECC 6.1 U8, Performance Manager, SMC and SPA. The question is do you need ECC ? It's a very "fat" and sophisticated product that can do so much more then just manage your Vmax.

HI,

SPA alone requires 4 GB of available physical memory. If you have plans to run ECC, SPA and SMC on the same server, make sure that you have enough RAM and compute to accomadate all. If your motto is just have a software for managing and monitoring Symmetrix then SPA and SMC are good.

Thanks,

Saurabh

Looking at the requirements, it appears the server we were planning to run everything on would not be able to handle it all (don't have enough memory, etc.) What we are looking at now is buying a server with enough CPU and memory to run as an ESX server with local storage, and run the various products (not ECC at first, but SCM/SPA and ESRS/Policy manager) in VM's so we can  have some flexability with upgrades or testing using another ESX server.

Do we have to worry much about storage space needed growing much over a few years, or will it top out after a while, and/or can we archive older performance data elsewhere? 

actually SPA requires 8G now.

Dynamox, which version requires 8 GB? If you break the total requirement of SPA, 1 GB is required by MySQL process and 1 Gb is for SPA service and rest 2 GB is for other things (I assume its for OS and other imp processes).

2.1 (includes OS requirements in the 8G)  ..actually SMC and SPA are "one" product now, one installer. You still need to get license separately for each application.

>For ESRS ..were you planning on putting your gateway VMs on different ESX servers ?

Yes, and not on storage backed by the array that it is the gateway to! We previously had it on VM's that were backed by storage from another array vendor, but as we are migrating everything to the VMAXen, we will need to have some way to ensure that EMC can still access the array if we are having issues.

We could consider having the gateway at another site in our metro area network, but would not want to worry in case of a disaster, or during a test where we cut the links (although I guess we could reroute the ESRS traffic to the local array prior to the test).

I've been running SPA for about 7-8 months now, collecting stats for 2 Vmax, 1 DMX3 and 1 DMX4. Right now mysql db size is around 30GB.  Now if you use performance manager and ECC, disk requirements will depend on WLA (workload analyzer) retention/collection policies. So same number of arrays with default retention/collection policies, in 1 year i have about 40G worth of performance data.

For ESRS ..were you planning on putting your gateway VMs on different ESX servers ? You should otherwise you have a single point of failure and i can tell you from experience when your site drops offline in ESRS ..you start getting a lot of calls from EMC because they think your systems all went away .

We could consider having the gateway at another site in our metro area network, but would not want to worry in case of a disaster, or during a test where we cut the links (although I guess we could reroute the ESRS traffic to the local array prior to the test).

ESRS traffic is all IP ..not sure what you mean by "reroute to local array".

mccready wrote: >For ESRS ..were you planning on putting your gateway VMs on different ESX servers ? Yes, and not on storage backed by the array that it is the gateway to! We previously had it on VM's that were backed by storage from another array vendor, but as we are migrating everything to the VMAXen, we will need to have some way to ensure that EMC can still access the array if we are having issues.

yep,  my gateways are running on local storage as well. I actually got some older DL 360 G4 and running ESXi (the free version).  I have two sites as well (metro), one site has Policy manager and two VM gateways, second site has just the two gateways that report to Policy Manager at the other site. I am ok with Policy Manager going away because my gateways cache authentication info so EMC could still get in (my policy is set to to auto-allow EMC to connect). I do backup Policy Manager to the other site so if i had a true disaster i can rebuild Policy manager relatively fast and restore from backups. I just did not want to dedicate another VM for Policy manager.

If we are worrired about the ESRS gateway running on the storage of the VMAX array it is the gateway to, we could have it run from another datacenter in our metro area on a VM backed by storage from an  array there.

However, when running in DR mode, we would still have to have a local to the datacenter  ESRS gateway as our network connections to the other datacenter may be down, which is why we are looking to run it on server that does not use SAN storage. So that is why we are looking at running the SMC, ESRS and possibly ECC tools on a single, well configured, ESX server with local storage. Anyone do that?

let me make sure i have it right. So at each site you will have an ESX server that will host two VMs (let's just talk about ESRS setup). One VM will monitor local arrays and another VM will monitor arrays at remote site ?

ok, so

Site 1:

how many VMs that are "gateways" ?

i assume they will report to local Policy Manager ?

Site 2:

how many VMs that are "gateways" ?

what Policy Manager will they report to ? (Site 1 or DR) ?

DR site:

how many VMs that are "gateways" ?

i assume they will report to local Policy Manager ?

Not exactly; our desired configuration (at this point) is:

Primary site 1: (VMAX in production)

- ESX server running ESRS, failover Policy Manager, SMC, SPA and backup ControlCenter

- SYMCLI management server (used for scripts and provisioning)

Primary site 2: (VMAX to be installed this year)

- ESX server running ESRS, SMC, SPA

DR site: (VMAX mainly used for DR)

- ControlCenter server with Storage Scope and WLA (present on dedicated physical server)

- ESX server running  ESRS,  Policy Manager, SMC and SPA

- SYMCLI management servers (used for scripts and provisioning)

The ESX server will hopefully have a very robust configuration to be able to run and hold all needed resources.

For all sites, we assume all would work from one policy manager (in this case, DR), with a failover policy manager configured. We're only planning to have one ESRS gateway per site - I'm not sure why we would consider more, as we will only have one VMAX per site..