December 21st, 2011
Ron Lee Gartner VP and Distinguished Analyst Tom Bittman’s opening comment, “Welcome to the peak of the Cloud hype cycle”, was very appropriate. Tom covers virtualization and cloud computing environments, and questions like: What to do about them, how to implement them and how to manage them dominated the session topics. Gartner Datacenter is an interesting conference, because it is where the IT community comes to get an update on what is going on, find out what’s new and to see what everyone else is up to.
One session I particularly enjoyed was “Private Cloud Computing at Bank of America – One Year Later”, presented by Bank of America. The room was full, and the audience was engaged and generating questions on an insightful success story. A year ago, Prentice Dees from Bank of America presented on the company’s plans to create a private cloud. . This year, he came back to declare victory and report on their success. They had successfully built and were operating a basic private cloud that could deploy VMs with applications installed automatically in 3 hours or less. This was a compression down from 60 days using the old manual processes. Prentice did admit that this was only 1/3 of the problem. Figuring out what to do and how to use it took just as long as before and were still the dominant costs of getting a new service deployed in a business unit. It was a good reminder that, in spite of all the technology, the key factors for success are still people and process.
I think I have some reading and listening to do so I can better understand what customers need from cloud computing. I see hints of things to come in fabric computing and the DevOps methodologies. As always things are moving faster and the abstractions are becoming greater, but sometimes customers need to be able to dive deep into the infrastructure and see what is going on. That is what Virtual Instruments is here for.
November 29th, 2011
Ron Lee Hi! You’ll be seeing more of me throughout the coming weeks and months, and I thought I would start out with an introduction. My name is Ron Lee and I am the new Technical Marketing Manager at Virtual Instruments. This marks the end of my first month at Virtual Instruments and it has been great fun.
Before joining the Virtual Instruments team, I was with VMware helping virtualization dramatically change data center management. VMware has done a fantastic job of making virtualization happen and making it possible to create the cloud data centers we are all working on. I came to Virtual Instruments to help the team begin creating the tools, metrics and practices needed to deliver apps in private cloud data centers.
In this role, I have been tasked with exploring and explaining a lot of the details of how Virtual Instruments products and services help companies resolve mission-critical virtual infrastructure issues. One of the first things I am doing is looking at our customer use cases, so you will mostly see me on the SAN Best Practices blog with a strong emphasis on application delivery. Application delivery is closely tied to the company’s founding mission, and is a hot topic for our business customers.
In addition, earlier this month, I was able to attend the Storage Decisions show in San Francisco with the Virtual Instruments team and spent a majority of the day in the company’s booth. The nice thing about Storage Decisions is that it attracts a qualified audience and brings in people from around the San Francisco Bay Area. Since this was my first show with Virtual Instruments, I was looking forward to hearing about customer challenges at the storage level for data centers, and I heard some interesting stories. For instance, I heard numbers like 180 switches and thousands of fibre channel port in a single data center. I was thinking of how long it would take to just visit each switch or array to look for issues before they caused an outage. It is challenges like this that help make it clear how necessary Virtual Instruments is for so many of our customers, and I look forward to sharing more tips and updates on how we are helping to enable businesses migrate to the private cloud. Have a SAN virtualization story you’d like to share? Drop me a line on the blog.
May 4th, 2010
blogadmin As we’ve been discussing for the last 18 months, and several years before that with Finisar, tapping is really becoming a best practice in the fibre channel storage networking, much like it’s been a best practice for IP networking for many decades.
Tapping your storage network is becoming a best practice because it enables non-disruptive diagnostic monitoring of the storage network infrastructure at any time. The beauty of tapping is that you’re watching a mirror image copy of the data from a passive, non-powered device. So it’s safe. And by being able to access the physical layer in a non-disruptive way, you can determine, preventatively, where you’ve got potential issues, and resolve them. To be able to easily remedy issues, and their root cause, has led to a tremendous success in the adoption of TAPs (Traffic Access Points).
That said, tapping storage networks has often been a new and additional step in the labor chain of building out net new SAN infrastructure. So when you deploy new hardware, you plan the TAP link and the location of the TAP, and the racks, etc., in addition to many other physical layer connections.
After a lot of discussion with a lot of customers, we’ve said, well, you’re going to have patch panels, and you’re going to set up new physical layer infrastructure for new equipment anyway, so why don’t you tap at that time?
Then we came to the conclusion that you can actually tap and patch at the same time. They’re both passive components, which consume no power and have no impact on performance. The process of installation is also similar. It’s the cable guy making connections. Basically, what they’re doing is they’re pulling fibre and making connections. And most of the cost is in labor.
So we said, why don’t we combine the two? And we did, and we call it a tap patch panel system, which we announced as SANInsight TAP Patch Panel System. It’s a patch panel and tap combined in one system that allows you to make your patch panel connections like you normally would, and automatically get a tap, in effect, for a very nominal cost. But most significantly, with a big labor savings whenever you’re building out net new infrastructures.
So far, the response from our customers has been very positive, and we expect that this simple idea will help further enable tapping as a best practice.
April 23rd, 2010
blogadmin Recently, we’ve done a number of SAN optimization studies for customers, and the general pattern seems to be consistent, in that typically 95% of the SAN ports are utilized less than 5% of the time. The interesting response to that finding we’ve gotten from some customers has been “we’ve been told we need that number of connections for any-to-any connectivity”. What is really over-provisioning has been presumed to be Best Practice. But in fact what’s happening is that people are paying a 95% premium on port purchases for the apparent benefit of any-to-any connectivity.
We asked one of our Solutions Consultants, Andrew Benrey, to take a typical customer configuration, this one was from a mutual fund company, and show them how to reconfigure it with fewer ports that were more highly utilized, while still maintaining any-to-any connectivity.
The benefit of port consolidation is that it can positively impact the entire system. Because you need fewer ports, you also need fewer SFPs (small form factor pluggable optical transceivers) which, according to some of our friends at Brocade, is the number one cause of SAN outages. So you reduce the number of assets you need to purchase, maintain, and monitor. You’ve got fewer things that can go wrong, and fewer places to look when something does go wrong. It decreases capital spending while simultaneously increasing operational efficiency. People are getting overwhelmed with number and complexity of assets under management. This is a simple way to reduce complexity and cost, while actually improving availability.
Consolidating the SAN is very similar to the VMware story of consolidating servers. Consolidation saves both administration time and money and increases overall reliability as there are fewer components to fail. Nearly all of the customers we have spoken to are deploying VMware for its obvious financial benefits. Once they understand the savings opportunity around their SANs and storage, they get even more excited with this “newly found money”.
April 8th, 2010
blogadmin We recently did a SAN audit for a large customer with a critical e-commerce application. They had purchased the best-of-the-best in terms of storage, switches and servers, in a highly virtualized environment. We came in and performed a SAN audit, and found that in terms of potential issues and problems, they actually had a pretty clean SAN, which was pretty impressive.
We also found out that they typically had response time numbers around 40 milliseconds, and sometimes peaked at 400 milliseconds. Although, this may appear fast, to put this in perspective, the Oracle recommendation for SAN infrastructure is 8 milliseconds. And we often see customers who’ve tuned with VirtualWisdom in the low single-digit millisecond range. This customer, like many others, tended to speculate about what it takes to improve performance – as they had no means to directly measure this before speaking with us
When it comes to performance tuning, we don’t like to speculate; we like to measure, with true cause and effect analysis. In the case of this customer, it was something as simple as adjusting their queue depth settings. Queue depths are kind of like metered on-ramps to the freeway, which allow traffic from the servers through the HBAs on to the Fibre Channel SAN. These folks spent $40M on an infrastructure procurement, for servers, SANs, HBAs, etc. And with a relatively modest investment in performance tuning capability, they found that they can substantially improve their performance. So the moral of the story is that next time you spend big bucks and buy a Ferrari, spending a small amount to keep it tuned up is an obvious worthwhile investment.
Entries RSS