October 19th, 2007 by James Cerwinski
Categories: IT Admin, Virtualization
SMB Virtualization
The basic value proposition of virtualization is the same for an SMB as it is for a larger enterprise. An SMB is just as likely to deploy virtualization to consolidate servers, improve server utilization, improve disaster recovery, consolidate legacy applications and reduce server provisioning time. The SMB experiences the same type of benefits such as reduced cost of hardware, energy, space requirements and provisioning.
VMware has just released 3 new bundles specifically targeted to the SMB market. According to industry analyst firm the Yankee Group, virtualization deployments among SMBs is expected to double during the next two years. VMware is positioned to capture this growth with their free product VMware Server and now offering three competitively priced bundles.
October 17th, 2007 by James Cerwinski
Categories: IT Admin, Virtualization
Virtualization 2.0
Gartner’s top 10 strategic technologies for 2008 has been published by networkworld.com. These are the technologies that are potentially game changers for companies so it is advised that all IT people examine them. In today’s environment, you don’t want to be behind the innovation curve.
Number one is Green IT. Cost and environmental concerns are driving the public, companies and the government to raise the awareness and urgency for action in this area. The early result has been new designs for chips, software programs to measure and manage power usage, and new technologies such as virtualization. Almost every virtualization story that I have heard or read included a claim to have reduced power usage. It would be great if our innovation in this area resulted in a better environment and much lower oil prices.
Number 5 is Virtualization 2.0. This is a catch all term for the value offered by this technology beyond server consolidation. Think about the simplification of installation and movement. A good example is a university that has a need to build and tear down a laboratory computing environment every semester.
I suggest you read the full article.
October 12th, 2007 by James Cerwinski
Categories: IT Admin, Virtualization
Virtualization of servers saves power and space for M&T Bank
Tom Cronin from M&T Bank presented the following results on their virtualization deployment. I found Tom to be practical, knowledgeable, and interesting with a good sense of humor.
They have deployed 70 production ESX servers (2 & 4 CPU HP servers) with 1,000 guests across two data centers and multiple DMZs.
This produced a 7% reducing in new hw purchases vs. double digit growth in the past years. He now claims to have excess rack space and lowered his power use. The environment thanks M&T Bank for this. Just imagine the benefit to the environment and energy costs if we all could lower our energy use. As a side note, I installed compact fluorescent bulbs in my house this spring and lowered my electric bill by 15%.
His key current pain points are server sprawl and charge back.
He is in the process of pulling together a project to work the following management issues.
- Where physically is the server?
- Virtual center CPU usage is not 100% accurate
- Provide performance data to end users.
October 3rd, 2007 by James Cerwinski
Categories: IT Admin, Virtualization
Virtualization Roadmap
There are many players in the virtualization field and each produces multiple product releases a year. Keeping track of the product releases they produce is a full time job. Virtualization.info does a pretty good job of it. They maintain a Virtualization Roadmap of the releases planned by the top vendors - including Microsoft, Citrix, XenSource, VMware, and Virtual Iron.
September 27th, 2007 by James Cerwinski
Categories: Development & Testing, Asset Management, IT Admin, Virtualization, Network Monitoring
VMware Infrastructure Software Development Kit (SDK)
We have posted a number of articles that highlight virtualization’s high value proposition. Very few things in life provide such gain with no pain. As pointed out in our Georgia-Pacific article, one of virtualization’s main pain is in the area of IT operations management. Our particular area of expertise and interest is in IT Asset Management and Network Monitoring. We are excited about the opportunity to discover, track, and monitor those virtual machines as they are created and zipped around from host to host at lightning speed. The rapid creation and movement of systems presents new challenges for IT operations management systems. Correctly leveraging and implementing the right set of management APIs can make the job much easier.
Fortunately, we have one of the best operations system interface engineers in the industry. Rakesk Bisaria has designed and implemented many mission critical systems which inter-operate to provide seamless flow-through operations. He has submitted the following article for me to post.
By Rakesh Bisaria;
VMware Infrastructure - consists of a suite of products and features that allow creation of Data Center Operations Applications. Some of the building blocks are:
VMwareCenter – VMwareCenter is the central management application that automates and optimizes the management of VMware IT infrastructure.
VMware Hosts - Servers (ESX Server. VMware Server) that host Virtual Machines (VM).
VMware High Availability (HA) – High Availability or fail-over capability for Virtual Machines that eliminates the need for a dedicated standby server and allows a VM to be available across the virtualized IT infrastructure.
VMware Distributed Resource Scheduler (DRS) – Dynamic load balancing and resource distribution by moving Virtual Machines (VMotion) to realize optimal performance across the virtualized environment.
VMware VMotion – Automatic move of VM from one physical server to another with no impact on the end users.
VI SDK Features & Services
The VI SDK supports the following Host Management and Virtualized Environment Management functions:
Host Management Operations - Following are some host management operations supported by the SDK VI API:
- Reboot or Shutdown a host
- Connect or disconnect a host from VirtualCenter
- Create or remove datastore from a host
- Configure the networking and storage systems attached to a host
Virtualized Environment Management Operations - Following are some Virtualized Environment management operations supported by the SDK VI API:
- Virtual Machine Creation and Deletion
- Virtual Machine Provisioning
- Virtual Machine Inventory
- Virtual Machine Migration (VMotion)
- Virtual Machine Performance Data
- Distributed Resource Schedule (DRS) Services
- High Availability Services
VMware SDK Development Environment – Following are the VM SDK Development Environment components:
· VMware SDK provided WSDL file
· Web Service Development Environment: Examples: Apache Axis, Microsoft Visual Studio .Net and .Net Framework
Over the next few weeks I will publish additional thoughts on this topic. In the mean time please provide your thoughts and comments.
September 26th, 2007 by J
Categories: Development & Testing, Virtualization, CommandCenter NOC, Network Monitoring
Virtual appliances from the software developers point of view
My name is Jason and I am one of the software engineers on the team. I joined the team after Raritan bought Oculan. My job is to create new features, improve performance, maintain and repair older functions. While I cannot take credit for much of the code base, I can tell you that I am happy to work on it. I work with some of the most knowledgeable and talented network management software engineers around - Including a co-founder of OpenNMS and three of the original lead OpenNMS developers.
When I came to Raritan, I was excited to find a team that was doing network management in the Raleigh area. On top of getting to work on network management Raritan was developing network management software within a standalone appliance. This appliance strategy was one that I had suggested at previous positions.
On the development side the appliance model quickly increases simplicity and decreases overhead. We write code for one database, one operating system, and one web server. We do not have to worry about competing for server ports, database resources, IO time, or OS prerequisites with other applications. This means we have more time to do the fun stuff, coding new features, exploring new products, and improving performance. By minimizing the time spend doing “plumbing”, we get more time to find new ways to improve your life. Yeah, I’m a geek. I like this stuff. I would always prefer to solve real problems and provide real solutions over spending time fixing issues that could be solved strategically.
On the user side, the appliance model provides software that is focused on network management and not interoperability matrices. An appliance does not just provide increased functionality for the buck. It provides greater security, more reliable performance, improved serviceability and easier upgrades.
A network management device gathers information that helps you understand your network. This is information an intruder can use to understand their own options for compromising a network. You need it to be secure. By taking charge of the operating system, we are able to maintain it. We aren’t going to require a Windows Admin to go out and buy an HP-UX or BSD box. You aren’t going to want to be responsible for securing an operating system you’ve never used before. We are going to take care of administrating it for you. Plus, an appliance is not going to have anything more than what is required to do its job properly. We harden the operating system and we do not open up a potential vulnerability that is unnecessary for our operation.
The appliance applications are the only user of resources on an appliance. By being the sole resource user an appliance application does not need a complex installation process to determine what resources are already used. Nor does it need a complicated start-up process to see what resources were taken by another application during downtime. With network management, the application is timing sensitive. You want to know about a potential problem or a real outage as soon as possible, so you can do preventative maintenance or in the case of an outage get folks back online sooner.
When it comes to serviceability, we are able to service your equipment more completely because we can securely control your whole operating system (with your express permission). We can recreate a problem you have been seeing and we can work to fix it immediately and gain greater insight on the cause. This insight allows us to write software to prevent a recurrence.
With all these positives, we have found some disadvantages to a hardware appliance. A hardware appliance takes up a whole 1U in your rack space. It would be eating space and power that might go to multiple virtual machines. Plus, if there is a hardware failure (they are rare, but they happen), you have to RMA the hardware back to the vendor. That is time lost without network management. Sure, anyone with good customer service is going to work to get you back running asap, but that’s time lost when you could’ve thrown in some spare parts and continued with life.
Raritan has recognized that a virtualized environment provides a number of benefits to our users. We understand that our job is to make your life easier. Just as I prefer a uniform installation base to code against, you prefer uniform hardware to manage. You can buy groups of identical servers and keep some spare parts on hand. Plus with all these spare parts, you can fix the hardware quickly and take advantage of the easier backup and restore processes that come with VMware. You get more time to do the fun stuff. By being prepared, you are able to recover from disaster with greater efficacy.
What about those other benefits of being a standalone appliance? What about the increased functionality, the greater security, reliable performance, serviceability, and upgrades? Well, we are a virtual appliance. We still write software that works on the same operating system as before. We have added a few new functions to our product creation process to handle the idea of being inside VMware instead of being inside a piece of hardware. So, no loss in functionality. VMware provides a secure location for us to run, we are not sharing a file-system or data with any other applications. No other virtual machines can access your network management data. Your performance is controlled just as you control your other VMware virtual hosts. You can provide limits to our virtual appliance and to other virtual hosts to guarantee resources for network management. Our serviceability and upgrade-ability have not not changed between our hardware and full featured virtual hardware offerings.
We will be posting about network management, security issues, virtualization, and whatever else we think you will find relevant. We hope you find our contributions informative. We know this is yet another corporate blog, and our content will be the only thing that can make it stand out. We took the same approach with our product. The content of our free download is built to stand out too.
Thanks for reading,
J
September 25th, 2007 by J
Categories: Development & Testing, Virtualization
Raleigh Area VMware User Group
On Thursday, a few of us developers went to Duke Clinical Research Institute to participate in the Raleigh VMware User Group. Our team not only develops a VMware virtual appliance, but we are also VMware users. We use VMware in our test environment to rapidly build very large test networks and Matt our “IT guy” is working on a VMware project. Using the product has improved our testing and saved us money.
I guess Raleigh is still a small town. One of the presenters is a guy I knew from the dorm and a friend of the family sat right down beside me. The really cool part of these meetings is the opportunity to connect with friends and others with a similar interest to share and learn about virtualization.
The talks were on virtual storage and automated tape backups within VMware. They were focused solutions provided by a couple of VMware partners. It was interesting to note that they took care to have a very similar look and feel to VMware. This targeted design makes it seem like the storage and backup industries are taking advantage of the rapid growth in VMware adoption.
Attending the meeting was a good investment of my time. I plan to continue to invest time to share and learn.
Jason
September 22nd, 2007 by James Cerwinski
Categories: IT Admin, Virtualization
Server Virtualization Project Spend Data
Server virtualization is a catalyst for technology refresh. It is estimated by VMware(r) that every $1 spent on VMware products generates the following additional spending:
- Server Equipment - $2-4
- Storage - $3-5
- VAR SI services - $2-3
- Infrastructure Management - $1-5
- Infrastructure Availability - $1-5
Share your server virtualization project spend ratios.
September 20th, 2007 by James Cerwinski
Categories: Asset Management, IT Admin, Virtualization
Georgia-Pacific’s Virtualization Expirience
Brad Wagner from Georgia-Pacific presented about their VMware(r) virtualization program. The following are the highlights:
Original business case was approved based upon server consolidation at a 5 to 1 ratio.
Results:
- Containment(rapid upgrade & deployment) has proven to be a valuable benefit not listed in the original business case
- Actual consolidation is at a 10 to 1 ratio. 100 ESX hosts supporting over 1000 guest machines
- G-P adds 100 new OSs a year - 80% are deployed as virtual machines. 20% require new hw. Long term goal is to implement 100% of wintel requests as virtual machines.
- Cap spending has dropped
- Provisioning time has been reduced dramatically
The main constraint is managing the rapidly growing environment. The following are some of the pain points mentioned by Brad(not in any particular order).
- Trouble shooting application performance
- Storage consumption planning
- Asset management
- Manage charge back
- Process automation
- Performance monitoring
- Change management
- Patch and update
Please share your virtualization deployment experience.
September 18th, 2007 by James Cerwinski
Categories: IT Admin, Virtualization, Network Monitoring
VMware(r) Guest Machine Performance Monitoring Issues
You should be aware of the following two points when monitoring performance of a guest machine. These two points may make traditional measurements inaccurate at certain times. The recommendation is to look at the host system first if you suspect a performance issue.
OS time keeping by a guest machine is sometimes inaccurate. An OS tracks time by the deliver of interrupts. Interrupts may not be delivered to the guest at a constant rate due to CPU over commitment.
Guest machines report total memory allocated by the user. However, memory is actually allocated to the guest by the hypervisor on a requested basis.
Let me know if you have experienced this or other system monitoring issues on a VMware(R) guest machine.
References
Understanding Performance in Virtualized Environment Presentation; by Krishna Raj Raja, Sr. MTS VMware
