Disk Enclosures
1748180 Members
3886 Online
108759 Solutions
New Discussion юеВ

Re: EVA performance tools

 
SOLVED
Go to solution
Tom O'Toole
Respected Contributor

EVA performance tools

Can anybody say if there will be performance tools available for the EVA coming out in the near or distant future? Ones measuring the utilization of each component of the system, like VTDPY? There have been numerous times where we could have made use of such tools. Thanks.

-Tom O'Toole
Can you imagine if we used PCs to manage our enterprise systems? ... oops.
12 REPLIES 12
Fernando Bandeira
Frequent Advisor

Re: EVA performance tools

Hi Tom,

Right now Storage Optimizer (part of OVSAM) is the only publicly available performance monitoring tool for the EVA.

I have not seen any external documentation mentioning any other tools in the future.

Regards,

Fernando
Alzhy
Honored Contributor

Re: EVA performance tools

Actually OVSAM relys on HP MeasureWare Agents to track EVA LUN performance on the servers - not the EVA themselves.

And it is not even real-time!

Hakuna Matata.
Fernando Bandeira
Frequent Advisor

Re: EVA performance tools

That is only one aspect of Optimizer functionality. The DPI for the EVA allows performance collection at a physical level as well.
Tom O'Toole
Respected Contributor

Re: EVA performance tools

Fernando,

Can you elaborate on what information this provides? Is this an optional part of ov/SAM, I haven't noticed it? Thanks!
Can you imagine if we used PCs to manage our enterprise systems? ... oops.
Alzhy
Honored Contributor

Re: EVA performance tools

Same here... AFAIK... there is no performance monitoring tool specifically for the EVA.. it's on a per LUN basis via the server's monitoring tools OR the SAN switch.

DPI? Could you please elaborate?
Hakuna Matata.
Uwe Zessin
Honored Contributor

Re: EVA performance tools

DPI = Device Plug-In

It knows about the features of that particular device it is written for.
.
Stephen Kebbell
Honored Contributor
Solution

Re: EVA performance tools

Hi Tom,

Storage Optimizer is a part of OVSAM, but when you buy an EVA only licenses for Storage Node Manager are included (also called Storage Operations Manager). You get a 60-day "Instant-On" for all components of OVSAM (Node Manager, Builder, Accountant, Optimizer, Allocater). If you want to use Optimizer after the 60 days are up, you need to buy a license for it.
When the EVA DPI is installed (it's not installed by default with OVSAM 3.1), and host agents are deployed on hosts connected to the EVA, you can start data collection for the EVA. As mentioned already, the data is not real-time. It collects data at intervals (I believe you can set the interval). You have to specify what metrics to collect - the metrics are listed in the DPI release notes.
http://www.openview.hp.com/products/dpi/install_guides/HP_HSVdpi_03_10_C.htm

Regards,
Stephen
Fernando Bandeira
Frequent Advisor

Re: EVA performance tools

Hi Tom,

As Stephen already mentioned, there are various metrics that performance information can be collected for. Please note also that Optimizer is not only for the EVA - it monitors switches, HBA, ports, other storage systems, etc. Yes, it is not real time, but you can set the intervals. The advantage is that you can the keep this informaiton in the database, which can be used for a variety of purposes, such as planning, baselining, etc. Once you have been collecting information for a while, a baseline for the metrics (not all of them) can be established, whcih you can then set thresholds against, which means that should performance numbers fall outside of those thresholds, you can use OVSAMs notification capability to be notified.

Regards,

Fernando
Tom O'Toole
Respected Contributor

Re: EVA performance tools

Thanks...

These performance metrics are (much) better than nothing, although I would also like to see some things like - %CPU utilization within controller, top utilization processes - like VTDPY. Otherwise how does one know where the bottlenecks are?

To give an example, I tested single jobs on VRAID5 vdisks and VRAID1 to get timing information. VRAID1 was 15% slower in this testing. However when running many simultaneous jobs, the VRAID1 performed unexpectedly faster. Could it be that the extra CPU needed by raid5 created a controller CPU bottleneck with more than N jobs? Who knows? Because subsystem utilization data isn't available.


[Am I crazy, or is it a bit unreasonable to have to pay extra to buy another license to get performance information? After all, the customer wass already forced to purchase a microsoft management station. What if they don't want or need fancy, buzzword compliant, Gartner certified, enterprise enabled, management framework, vendor committed, best-of-breed, director-class, value-added software (and certainly don't want to pay for it), they just want access to the info?]
Can you imagine if we used PCs to manage our enterprise systems? ... oops.