How many virtual machines can I run on one server?

The number of virtual machines that can be run on one server actually depends on the physical box configuration—the memory, processor speed etc.—and resource demands of the applications on each virtual server. For example with CPU-intensive workloads, memory becomes a significant factor in performance and hence in overall VM capacity. In fact with very powerful boxes, you usually run out of memory and disk I/O before all the CPU power is exhausted.

A more powerful server and lower application resource demand means you can host a greater number of virtual machines. Conversely, lower box power and higher application resource demand mean fewer virtual machines per physical server.

That’s why at Mosaic as a first step in any virtualization we give our customers an informal VM Capacity Discovery session. We can give a first level assessment of what level of consolidation to expect.

(0) Leave a Comment

Virtualization: The Numbers Don’t Lie – ROI Still Looks Good in a Down Economy

ROIThe stock market is tanking. AIG executives are spending almost a half million dollars for a luxury spa retreat right after they got a chunk of change from the $700 billion bailout. Companies around the country are tightening their belts and cutting back on expenditures. All across America, budgets for new IT purchases is getting reevaluated.

So why is this a great time for virtualization projects? Simply because the ROI on virtualizing your datacenter can be less than a year, saving you money on your bottom line THIS budget year! Depending on the size of your datacenter, and your new server refresh cycle, you may be able to repurpose this year’s budget to a virtualization project and actually decrease your bottom line.

I know, that sounds almost impossible, but let me show you how it works. Assuming that you have a datacenter that is populated with … say 50 servers. They can be a mix of Windows, Linux, or other x86-based hosts. Another assumption is that of the 50 servers, you are retaining them 4 years, and replacing them on a regular cycle, at probably 12 new servers per year. After memory, drives, etc, you are looking at probably $5,000 each for a ballpark number of $60k per year for new server purchases.

If you were to virtualize your datacenter with VMware, you should be able to migrate 40 of your existing servers to VMs as part of the project. In an ideal world, we could virtualize them all. However there may be a few servers that require unsupported expansion cards, are being outsourced, or are true ‘high performers’ that are not suitable virtualization candidates. A conservative estimate would be that we could support 40 virtual machines within a 3 host ESX infrastructure. It would be very feasible to re-purpose an existing server as your Virtual Center management server. Therefore, rather than purchase 12 new servers for this year, you would only require 3 new servers.

Continue reading “Virtualization: The Numbers Don’t Lie – ROI Still Looks Good in a Down Economy” »

(0) Leave a Comment

HP Ignites iSCSI Wars

HPOver the past few years, iSCSI has seen a slow transition from a maverick (thank you Sarah Palin and Tina Fey for keeping this word at the top of my brain) low-cost offering for block level storage into a mainstream solution. The maturity of this offering has met or exceeded all expectations. Dell’s purchase of EqualLogic last year was a clear signal that iSCSI is becoming a major player in the storage game. The simple fact that the Dell|EqualLogic line is their fastest growing storage solution, and Dell is putting the full weight of their marketing arm behind the product shows the versatility that iSCSI offers. Being the only major player with a custom iSCSI solution gave Dell a distinct advantage over the competition.

With the acquisition of Lefthand Networks by HP last week, the gauntlet was thrown down. Dell had a year’s head start, but there will now be some competition in the marketplace. HP was well placed in the FC market, with their MSA and EVA products, but were missing the iSCSI boat. Let’s be honest, the MS Unified Storage Server is NOT a viable solution. Lefthand fits nicely into their product line, and will be supported well by HP’s storage engineering team. In the first two quarters, there will most likely be a bit of a marketing blitz, and look to see some competitive pricing as well. After that, it will be interesting to see how it integrates into the rest of HP’s product line.

Meanwhile, I don’t expect that the Dell|EqualLogic team will be sitting on their laurels. Look for improvements to the feature set over the coming months to differentiate their product from Lefthand.

(0) Leave a Comment

VMWare and Disaster Recovery

Disaster Recovery PlanDisaster recovery has been a hot topic for the past few years. Disaster recovery was thrust into the mainstream media after Katrina hit New Orleans and devastated that part of the country. Many people started thinking about disaster recovery afterward and started to look at the solutions available in the marketplace. Hosted disaster recovery is available, but can be cost prohibitive for someone in the SMB market. Standalone servers in a secondary site can be problematic, as it introduces OS and application issues by running on different hardware. However, if a company has virtualized with VMWare, they are already on the track to a comprehensive disaster recover solution.

One of the most dramatic changes that occurs when a server is virtualized is that the server becomes compartmentalized. Basically, all of the files, configuration information, data, and drivers that make up a server in your enterprise are rolled into just a few files. There is a configuration file that tells the ESX server how much memory, drive space, network cards are in the server. Additionally, there is a “vmdk” file, which acts as the virtual hard drive. All OS, server config, applications and data that would reside on the a physical server’s hard drive is stored here in a single large file. Take into account that virtual servers use “virtual hardware”, which is platform independent and can run on any ESX installation. When your servers can be reduced to just a few files, it opens up several possiblilites. Files can be copied. Files can be backed up. Files can be replicated.

In all practical senses, a virtual machine (VM), has been reduced to simply a few files. Looking at this from a disaster recovery perspective, backing up and recovering a server has been simplified dramatically. Where once you needed to use a backup agent to perform a full backup, copy or clone the data to a second site, and have identical hardware available to restore the files to, you now have reduced the entire process to moving a few files. If you need to restore a VM from these files, any ESX installation will do regardless of if it is a Dell, HP, or IBM server on the back end. You can restore a VM running on an HP blade to ESX installed on a Dell rack-mounted server. The difficult part is copying the VM files from your production environment to the DR site, which really isn’t that challenging.

Technically, how does VMWare integrate into a comprehensive Disaster Recovery/Business Continuity plan? There are several options available, ranging from manual replication to using VMWare’s Site Recovery Manager product to develop the plan within your company’s requirements.

I think that this is enough for now. All I ask is that when you think of Disaster Recovery and Business Continuity, think of incorporating VMWare into your solution.

(0) Leave a Comment

News from VMWorld 2008: Update


VMWorld 2008, Las Vegas. As I stand in line, at least 200 people deep to get on an escalator to take me to the fourth floor of the Sands Expo Center for my next session, I am amazed at the crowd here at VMWorld. This is my fourth one, and I have seen it grow from 3500 at Mandalay Bay to the current estimate at over 14,000 attendees. I finally get on the escalator and say something like ” I can’t believe there are so many people here”. The guy behind me calmly says “Yeah. If a bomb went off here, everything around the world would stop. There would be nobody left to keep the servers running.”

While it was funny and everyone around us laughed, it had a ring of truth to it. Virtualization has become so pervasive in the IT space now that we have come to rely on it in almost every aspect of our datacenter. A full 100% of the Fortune 100 are using VMware in production, and 98% of the Fortune 500 use their products. With all of the top engineers in the field in one place, a disaster would surely cripple IT innovation for years to come.

Aside from that grim concept, the conference has been fantastic. With the new announcements surrounding vStorage, vNetworks, and the vCloud initiative, it is obvious to all that VMware is not sitting back and enjoying their lead in the Virtualization space. Every time their competitors start getting closer to what they are doing, they take a huge leap ahead and push virtualization to new heights. I am amazed every year that they continue to innovate, and look forward to getting my hands on the new technology.

(0) Leave a Comment

News from VMWorld 2008: Day 1

VMworld 2008One of the new features that VMware is touting for their new Datacenter Operating System is the ability to package VMs into a vApp.  In a nutshell, if you have a group of VMs that are working in conjunction to support an application platform, such as RubyOnRails, Python, .NET, or other closely integrated VMs to support a single application, you can package them all into a “vApp”.  This vApp can be defined by SLAs, resource pool requirements, or other business rules, and be monitored by vCenter’s upcoming AppSpeed module.  Once this vApp package has been defined, it can then be shared or bundled in a way to be run on an external computing cloud, or vCloud.  There are several third party resources available to host customer vApps, enabling the virtual datacenter to move outside of the confines of physical datacenter.

This concept was shown to the audience at the keynote this morning as a live demo.  Despite the common pitfalls of all live demos, it performed flawlessly.  A vApp packaged and running in a virtual center was stressed, and based on performance SLAs, a replica vApp was pushed out to a service provider vCloud, and run in parallel to balance the load and satisfy the SLA.

(0) Leave a Comment

Tim Antonowicz
Tim Antonowicz
Senior Sales Engineer

Tim has over 15 years of experience in Enterprise storage and backup, Virtualization, Disaster Recovery, Messaging, and cross-platform integration.

Tim is nationally recognized in the fields of Virtualization, Disaster Recovery, and Security, frequently speaking at national conferences such as VMWorld, HP Technology Forum, EduCause, Gartner MES and FinSEC.

He is also a charter member of VMWare's SMB Customer Advisory Panel and sits on the New England VMWare User Group's Steering Committee.

Tim has a Bachelor of Science degree in Biology from the Worcester Polytechnic Institute and holds various industry certifications including the VMware Certified Professional, Microsoft Exchange 2003, and has been a Microsoft Certified Systems Engineer since 1998.

Show More | Hide More