if i have the target Sizes
Applications like CRM and ERP
and have number of VDI uses
dont have number of IOPS
how can i have estimated configuration based on that ?
I read that there are some tools which are used to simulate scenarios and give estimated iops , is that correct ?
There are certainly a number of load testing tools you can use to help get an idea of how to size the system, but they may not be incredibly accurate to your specific workload.
Is there a non-VDI deployed desktop that you can use to get a bit more intel on before swinging users over to the View environment? The more info you have on the requirements for your specific workload, the better.
i found the VSPEX sizing tool Helpful
the problem is , I sent the needed details after I read some VDI Practices and testing that tool but seems that the questions are hard to be answered
yes it will not be very accurate I guess and will be based on assumptions
"Know your workload" is great advice-- but is really difficult to accomplish in most environments
You could always start with a small pilot group and get everything you need to monitor/capture the data set up in advance, then use that data to build from. Just be sure you choose "real users" that will test and put a realistic load on the systems.
As Steve said, it's quite complicated to provide you IOPS numbers since, by nature, every customer is different.
So, you have two choice, based on my experience:
- 1st : use a real assessment tool that will provide you real numbers, usage, eligibility to VDI based on agent based metrics you'll install on reference desktops. This can be done by acquiring licenses or buy the assessment service to a SI.
Some of these tools: Liquidware Labs Fit, Systrack, VMware...
- 2nd: work with assumptions based on customers experience and classify in many catoegories: i.e: light, medium, heavy, etc... IOPS in steady state (normal task) can be between 12 to 30 IOPS depending on template optimization (no aero, no unused services...), software resources, video, HD, etc...
Moreover, once you designed your profiles, usage, requirements, you need to set many things :
- concurrency % per profile
- disaster recovery scenario (RTO/RPO, degraded performances...)
- user data (where, replicated or backup)
- infra replication: synchronous or asynchronous,
all of these to take account into your storage architecture among entire infrastructure.
Feel free to ask me more details if needed