Managing VMware Infrastrucutre - time taken to do the job ?? advice and opinions needed please.

[big_vern]

guys / girls – some opinions needed on the amount of work needed to maintain and develop a VMware Infrastructure setup as follows:

9 HP Blades (BL 465 & 685) connected to an EVA 8000 (dual fabric connected)

Set up as 2 separate HA /DRS Clusters running ESX 3.5 Update 3

130 VMs, mainly running windows 2003 (full nightly VCB backups)

How many hours a week would you say, in your experience, would be needed to maintain, troubleshoot and develop (keep up to date with patches etc) an infrastructure of this size in a critical enterprise environment with 2,500 users (IE if something goes wrong then there will be serious questions asked). The job would mean ensuring DR ability and full uptime etc.

Be as conservative as you like, the duties include looking after the Hardware as well (with relevant HP support contracts), could you include brief reasoning if possible. The turnover of VMs is about 2 a week.

[MHAV]

First of all I wouldn't use a EVA8000 because it's not a full redudant storage system or do they make you belive it is one (it's an active passiv system)!? The only Active/Active Midrange Storage is the AMS by HDS.

9 HP Blades you mean 9 ESX Servers and you want to create 2 clusters and you only have 130 VM's?

Regards

Michael Haverbeck

If you find this information useful, please award points for "correct" or "helpful".

[meistermn]

Does the ams by hds has a virtual san like the metrocluster from netapp ?

I read this document and only found true copy, which is not metrocluster.

[azn2kew]

Sounds like you're looking to hire someone or contractors for estimation scope of work or something? Implementing this scope of work really depends how other team collaborate and if things already in placed such as DBA, Networking, Security,and application owners. Putting a new ESX cluster with 9 blades through HP EVA8000 and provision 130 VMs aren't that long but to get everything approved, planned, testings would take some time due to the fact you want someone to carefully architect so your environment would be reliable, scalable and performance wise.

If networking infrastructure and all hardware/software are ready to go this can be done in 1 week but varies depend how complexity it involved. I think if you have consultant company comes in and bid, they will bid more than 1 week for sure because they need to make money for the whole life cycle of the project. meetings, analysis, research, architect, planning, implementation, support way too much.

HDS and HP EVA8400 are pretty good solid SAN though! With this type of environment, I suggest you have a full time VMware Admin to take care of business on daily basis and other tasks if needed. Or have your technical staff train the products and learn from the contracts or transfer of knowledge would work.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[MHAV]

I'd agree to set it up and configure it, writting documentation and testing can take easy two week. On the daily basis I'd have a guy doing nothing else.

Regards

Michael Haverbeck

If you find this information useful, please award points for "correct" or "helpful".

[MHAV]

SAN it self is virtualization - I'm not sure if tha AMS has a metrocluster funtion/software like metrocluster.

What you need a extra virtualization with a SAN - there are no 2 instance SAN Virtualization Systems on the market.

You can use a Enterprise Class SAN and then have the complete system redundant and the you lose this protection with a software virtualization

Regards

Michael Haverbeck

If you find this information useful, please award points for "correct" or "helpful".

[big_vern]

Thanks guys, thanks for the answers which give me a starting point

one final part of the puzzle is how many man hours a week would you need to employ someone to look after the infrastructure once set up. (ie. checking backups, torubleshooting any performace issues, provisioning maybe a couple of new VMs a week, maintaining patch levels and looking after the hardware etc.)

Again be as conservative as you like. I'll award points after this bit.

Cheers..

[Texiwill · Accepted solution]

Hello,

one final part of the puzzle is how many man hours a week would you need to employ someone to look after the infrastructure once set up. (ie. checking backups, torubleshooting any performace issues, provisioning maybe a couple of new VMs a week, maintaining patch levels and looking after the hardware etc.)

For that size environment I would have a full time virtualization administrator that works closely with your storage, networking, server, and application teams to solve problems. Almost everything from the beginning will be considered a virtualization issue. You need a good plan on how to handle this. You will need a team comprised of members from other teams to handle all possible problems.

Again be as conservative as you like. I'll award points after this bit.

I would most likely aim for at least 2 people for full redundancy. The job does not come to a standstill when one is on vacation.

3-9 hosts can easily be managed by one person but let's do some math.....

150 VMs for start....

Let's say you need to VMotion all 150 VMs..... Say 30 seconds per VM, so 150 * 30 = 1.25 hours = average 2.5 hours for interruptions, etc.

Let's say you need to SVMotion all 150 VMs..... Say 1 GB/s with an average 24GB per VM so 3 work day to SVMotion everything.... Closer to 1 week. Its a disk issue.

Now add in the time to:

Debug Application Issues....

Debug Storage Issues....

Debug Network Issues....

Debug Guest OS Issues....

Debug ESX Issues.....

Debug Performance Issues....

And the time to add:

Reporting needed by Management....

Training Time....

New VM implementation Time....

Nerw hardware planning....

DR testing and planning...

etc. etc.

It is at least 1 full person if not 2.

Best regards,

Edward L. Haletky VMware Communities User Moderator, VMware vExpert 2009, Virtualization Practice Analyst[/url]
Now Available: 'VMware vSphere(TM) and Virtual Infrastructure Security: Securing the Virtual Environment'[/url]
Also available 'VMWare ESX Server in the Enterprise'[/url]
[url=http://www.astroarch.com/wiki/index.php/Blog_Roll]SearchVMware Pro[/url]|Blue Gears[/url]|Top Virtualization Security Links[/url]|Virtualization Security Round Table Podcast[/url]

[azn2kew]

Frankly speaking, if you concerns with employing full time admin, working with local consulting firm with SLA in place they can work out a plan for you that might be cheaper, but remember its $250/hour for a firm for joe blow elsewhere watch out their credentials and work ethics but doable. joe blow like me can do the trick too but only $110/hour and that's anytime 24/7 services but better of have someone locally incase. I would say spend 2 hours to check out things logs/systems/tasks per day and if they need time to provision new servers just add it on. 15-20 hours a week for part time guys to monitor is fine but full lifecycle administration tasks it requires full time admin period.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[heybuzzz]

If they're just managing what you listed I would say one, but if this same person is going to a bunch of meetings to talk about new VM's or P2V'ing old server then maybe 1 full and 1 part time. I was managing something similar to your specs but once we got heavy into "virtualize 1st" and server consolidation we added a second person.

You also have to think about having a backup person incase your #1 gets hit by the bread truck.