Resource allocation for Virtual Machines

Ever since I started transitioning our production systems to my ESX cluster, I’ve been fascinated how visible resource utilization has become.  Or to put it another way, how blind I was before.  I’ve also been interested to hear about the natural tendency of many Administrator’s to over allocate resources to their VM’s.  Why does it happen?  Who’s at fault?  From my humble perspective, it’s a party everyone has shown up to.

  • Developers & Technical writers
  • IT Administrators
  • Software Manufacturers
  • Politics

Developers & Technical Writers
Best practices and installation guides are usually written by Technical Writers for that Software Manufacturer.  They are provided information by whom else?  The Developers.  Someone on the Development team will take a look-see at their Task Manager or htop in Linux, maybe PerfMon if they have extra time.  They determine (with a less than thorough vetting process) what the requirement should be, and then pass it off to the Technical Writer.  Does the app really need two CPU’s or does that just indicate it’s capable of multithreading?  Or both?  …Or none of the above?  Developers are the group that seems to be most challenged at understanding the new paradigm of virtualization, yet are the one’s to decide what the requirements are.  Some barely know what it is, or dismiss it as nothing more than a cute toy that won’t work for their needs.  It’s pretty fun to show them otherwise, but frustrating to see their continued suspicions of the technology.

IT Administrators (yep, me included)
Take a look at any installation guide for your favorite (or least favorite) application or OS.  Resource minimums are still written for hardware based provisioning.   Most best practice guides outline memory and CPU requirements within the first few pages.  Going against recommendations on page 2 generally isn’t good IT karma.  It feels as counterintuitive as trying to breathe with your head under water.  Only through experience have I grown more comfortable with the practice.  It’s still tough though.

Software Manufacturers
Virtualization can be a sensitive matter to Software Manufactures.  Some would prefer that it doesn’t exist, and choose to document it and license it in that way.  Others will insist that resources are resources, and why would they ever recommend their server application can run with just 768MB of RAM and a single CPU core if there was even a remote possibility of it hurting performance.

Politics
Let’s face it.  How much is Microsoft going to dive into memory recommendations for an Exchange Server when their own virtualization solution does not support some of the advanced memory handling features that VMWare supports?  The answer is, they aren’t.  It’s too bad, because their products run so well in a virtual environment.  Politics can also come from within.  IT departments get coerced by management, project teams or departments, or are just worried about SLA’s of critical services.  They acquiesce to try to keep everyone happy.

What can be done about it.
Rehab for everyone involved.  Too ambitious?  Okay, let’s just try to improve Installation/Best Practices guides from the Software Manufactures.

  • Start with two or three sets of minimums for requirements.  Provisioning the application or OS on a physical machine, followed by provisioning on a VM accommodating a few different hypervisors.
  • Clearly state if the application is even capable of multithreading.  That would eliminate some confusion on whether you even want to consider two or more vCPU’s  on a VM.  I suspect many red faces would show up when software manufactures admit to their customers they haven’t designed their software to work with more than one core anyway.  But this simple step would help Administrators greatly.
  • For VM based installations, note the low threshold amount for RAM in which unnecessary amounts of disk paging will begin to occur.   While the desire is to allocate as little resources as needed, nobody wants disk thrashing to occur.
  • For physical servers, one may have a single server playing a dozen different roles.  Minimums sometimes assume this, and they will throw in a buffer to pad the minimums – just in case.  With a VM, it might be providing just a single role.  Acknowledge that this new approach exists, and adjust your requirements accordingly.

Wishful thinking perhaps, but it would be a start.  Imagine the uproar (and competition) that would occur if a software manufacturer actually spec’d a lower memory or CPU requirement when running under one hypervisor versus another?  …Now I’m really dreaming.

IT Administrators have some say in this too.  Simply put, the IT department is a service provider.  Your staff and the business model are your customers.  As a Virtualization Administrator, you have the ability to assert your expertise on provisioning systems to provide a service, and it work as efficiently as possible.  Let them define what the acceptance criteria for the need they have, and then you deal with how to make it happen.

Real World Numbers
There are many legitimate variables that make it difficult to give one size fits all recommendations on resource requirements.  This makes it difficult for those first starting out.  Rather than making suggestions, I decided I would just summarize some of my systems I have virtualized, and how the utilization rates are for a staff of about 50 people, 20 of them being Software Developers.  These are numbers pulled during business hours.  I do not want to imply that these are the best or most efficient settings.  In fact, many of them were  “first guess” settings that I plan on adjusting later.  They might offer you a point of reference for comparison, or help in your upcoming deployment.

Server/Function Avg % of RAM used Avg % of CPU used / occasional spike Comments
AD Domain (all roles) Controller, DNS, DHCP. 
Windows Server 2008 x64
1 vCPU, 2GB RAM
9% 2% / 15% So much for my DC’s working hard.  2GB is overkill for sure, and I will be adjusting all three of my DC’s RAM downward.  I thought the chattiness of DC’s was more of a burden than it really is.
Exchange 2007 Server (all roles)
Windows Server 2008 x64
1 vCPU, 2.5GB RAM
30% 50% / 80% Consistently our most taxed VM, but pleasantly surprised by how well this runs.
Print Server, AV server 
Windows Server 2008 x64
1 vCPU, 2GB RAM
18% 3% / 10% Sitting as a separate server only because I hate having application servers running as print servers.
Source Code Control Database Server
Windows Server 2003 x64
1 vCPU, 1GB RAM
14% 2% / 40% There were fears from our Dev Team that this was going to be inferior to our physical server, and they suggested the idea of assigning 2 vCPU’s “just in case.”  I said no.  They reported a 25% performance improvement compared to the physical server.  Eventually they might figure out the ol’ IT guy knows what he’s doing.
File Server
Windows Server 2008 x64
1 vCPU, 2GB RAM
8% 4% / 20% Low impact as expected.  Definitely a candidate to reduce resources.
Sharepoint Front End Server
Windows Server 2008 x64
1 vCPU, 2.5GB RAM
10% 15% / 30% Built up, but not fully deployed to everyone in the organization.
Sharepoint Back End/SQL Server
Windows Server 2008 x64
1 vCPU, 2.5GB RAM
9% 15% / 50% I will be keeping a close eye on this when it ramps up to full production use.  SharePoint farms are known to be hogs.  I’ll find out soon enough.
SQL Server for project tracking.
Windows Server 2003 x64
1 vCPU, 1.5GB RAM
12% 4% / 50% Lower than I would have thought.
Code compiling system
Windows XP x64
1 vCPU 1GB RAM
35% 5% / 100% Will spike to 100% CPU usage during compiling (20 min.).  Compilers allow for telling it how many cores to use.
Code compiling system
Ubuntu 8.10 LTS x64
1 vCPU 1GB RAM
35% 5% / 100% All Linux distros seem to naturally prepopulate more RAM than their Windows counterparts, at the benefit perhaps of doing less paging.
       

To complicate matters a bit, you might observe different behaviors on some OS’s (XP versus Vista/2008 versus Windows 7/2008R2, or SQL 2005 versus SQL 2008) in their willingness to pre populate RAM.  Give SQL 2008 4GB of RAM, and it will want to use it even if it isn’t doing much.   You might notice this when looking at relatively idle VM’s with different OS’s, where some have a smaller memory footprint than others.   At the time of this writing, none of my systems were running Windows 2008 R2, as it wasn’t supported on ESX 3.5 as I was deploying them.

Some of these numbers are a testament to ESX’s/vSphere’s superior memory management handling and CPU scheduling.  Memory ballooning, swapping, Transparent Page Sharing all contribute to pretty startling efficiency.

I have yet to virtualize my CRM, web, mail relay, and miscellaneous servers, so I do not have any good data yet for these types of systems.  Having just upgraded to vSphere in the last few days, this also clears the way for me to assign multiple vCPU’s to the code compiling machines (as many as 20 VM’s).  The compilers have switches that can toggle exactly how many cores end up being used, and our Development Team needs these builds compiled as fast as possible.  That will be a topic for another post.

Lessons Learned
I love having systems isolated to performing their intended function now.  Who wants Peachtree crashing their Email server anyway?  Those administrators working in the trenches know that a server that is serving up a single role is easy to manage, way more stable, and doesn’t cross contaminate other services.  In a virtual environment, it’s  worth any additional costs in OS licensing or overhead.

When the transition to virtualizing our infrastructure began, I thought our needs, and our circumstances would be different than they’ve proven to be.  Others claim extraordinary consolidation ratios with virtualization.  I believed we’d see huge improvements, but those numbers wouldn’t possibly apply to us, because (chests puffed out) we needed real power.  Well, I was wrong, and so far, we really are like everyone else.

Helpful links

Virtualization. Making it happen

 

It’s difficult to put into words how exciting, and how overwhelming the idea of moving to a virtualized infrastructure was for me.  In 12 months, I went from investigating solutions, to presenting our options to our senior management, onto the procurement process, followed by the design and implementation of the systems.  And finally, making the transition of our physical machines to a virtualized environment.

It has been an incredible amount of work, but equally as satisfying.  The pressure to produce results was even bigger than the investment itself.  With this particular project, I’ve taken away a few lessons I learned along the way, some of which had nothing to do with virtualization.  Rather than providing endless technical details on this post, I thought I’d share what I learned that has nothing to do with vswitches or CPU Utilization.

1.  The sell.  I never would have been able to achieve what I achieved without the  support of our Management Team.  I’m an IT guy, and do not have a gift of crafty PowerPoint slides, and fluid presentation skills.  But there was one slide that hit it out of the park for me.   It showed how much this crazy idea was going to cost, but more importantly, how it compared against what we were going to spend anyway under a traditional environment.  We had delayed server refreshing for a few years, and it was catching up to us.  Without even factoring in the projected growth of the company, the two lines intersected in less than one year.  I’m sure the dozen other slides helped support my proposal, but this one offered the clarity needed to get approval.

2.  Let go.  I tend to be self-reliant  and made a habit of leaning on my own skills to get things done.  At a smaller company, you get used to that.  Time simply didn’t allow for that approach to be taken on this project.  I needed help, and fast.  I felt very fortunate to establish a great working relationship with Mosaic Technologies.  They provided resources to me that gave me the knowledge I needed to make good purchasing decisions, then assisted with the high level design.  I had access to a few of the most knowledgeable folks in the industry to help me move forward on the project, minimizing the floundering on my part.  They also helped me with sorting out what could be done, versus real-world recommendations on deployment practices.  It didn’t excuse me from the learning that needed to occur, and making it happen, but rather, helped speed up the process, and apply a virtualization solution to our environment correctly.  There is no way I would have been able to do it in the time frame required without them.

3.  Ditch the notebook.  Consider the way you assemble what you’re learning.  I’ve never needed to gather as much information on a project as this.  I hated not knowing what I didn’t know. (take that Yogi Berra)  I was pouring through books, white papers, and blogs to give myself a crash course on a number of different subjects – all at the same time because they needed to work together.  Because of the enormity of the project, I decided from the outset that I needed to try something different.  This was the first project where I abandoned scratchpads and binders, highlighters (mostly) and printouts.  I documented ALL of my information in Microsoft OneNote.  This was a huge success, which I will describe more in another post.

4.  Tune into RSS feeds.  Virtualization was a great example of a topic that many smart people dedicate their entire focus towards, then are kind enough to post this information on their blogs.  Having feeds come right to your browser is the most efficient way to keep up on the content.  Every day I’d see my listing of feeds for a few dozen or so VMware related blogs I was keeping track of.  It was uncanny how timely, and how applicable some of the information posted was.  Not every bit of information could be unconditionally trusted, but hey, it’s the Internet.

5.  Understand the architecture.  Looking back, I spent an inordinate amount of time in the design phase.  Much of this was trying to fully understand what was being recommended to me by my resources at Mosaic, as well as other material, and how that compared to other environments.  At times, grass grew faster than I was moving on the project at the time, (exacerbated by other projects getting in the way) but I don’t regret my stubbornness to understand what was I was trying to absorb before moving forward.  We now have a scalable, robust system that helps avoid some of the common mistakes I see occur on user forums.

6.  Don’t be a renegade.  Learn from those who really know what they are doing, and choose proven technologies, while recognizing trends in the fast-moving virtualization industry.  For me there was a higher up front cost to this approach, but time didn’t allow for any experimentation.  It helped me settle on VMware ESX powered by Dell blades, running on a Dell/EqualLogic iSCSI SAN.  That is not a suggestion that a different, or lesser configuration will not work, but for me, it helped expedite my deployment.

7.  Just because you are a small shop, doesn’t mean you don’t have to think big.  Much of my design considerations surrounded planning for the future.  How the system could scale and change, and how to minimize the headaches with those changes.  I wanted my VLAN’s arranged logically, and address boundaries configured in a way that would make sense for growth.  For a company of about 50 employees/120 systems, I never had to deal with this very much.  Thanks to another good friend of mine whom I’d been corresponding with on a project a few months prior, I was able to get things started on the right foot.  I’ll tell you more about this in a later post.

The results of the project have exceeded my expectations.  It’s working even better than I anticipated, and has already proven it’s value when I had a hardware failure occur.  We’ve migrated over 20 of our production systems to the new environment, and will have about 20 more up online within about 6 months.  There is a tremendous amount of work yet to be completed, but the benefits are paying for themselves already.

Follow

Get every new post delivered to your Inbox.

Join 737 other followers