One of my passions is riding my bike (although you might not know it to look at my waistline). And I mean bicycle, not motorcycle (in fact, motorcycles should not be allowed to be called bikes, since there's no B, I, or K anywhere in the name. But that's an argument for another time).
I've been hurt a few times, but always on my mountain bike (which may be why I don't ride my mountain bike much anymore). I've always been much more of a roadie anyway. And I've never had a serious accident on my bike.
Until yesterday. I was out for a lunchtime ride on a glorious, 81-degree day here in north-central Maryland. I was on a fast downhill, probably going 35 MPH or more. Suddenly a cat starts crossing the road from the opposite side, not too far ahead of me. I see him/her, and gently apply brakes. I'm going too fast, however. The cat stops in the middle of the road. I literally shout out "Go one way or the other!" The devil cat then moves a little bit across the road, into my path -- then freezes, as if hypnotized. I swerve to the right to avoid him. But at the last second, this evil feline decides that that's the time to move. Right under my tires.
Bam.
Crash.
Over the handlebars; into the road. I slid on my left side for what had to be 8-10 feet. I distinctly felt my helmet smack, hard, against the pavement. I laid there for a few seconds, trying to figure out if I was dead or not. Deciding that I wasn't, I tried to get up (not good to be laying in the road with cars; fortunately, the roads I ride on for the most part are very lightly traveled.)
I was able to stand and walk to the shoulder. So far, so good. I could stand and walk. Next I moved my left shoulder, since all my pain was on my left side. It's extremely common for cyclists to break their collarbones when they do a header (head-over-handlebars). It felt OK. I didn't have any spots or diziness, which was another good sign; my next fear was of some kind of head trauma. Was it possible that all I had was some road rash?
I called my wife on my cell, which I always take with me on rides. The accident happened less than four miles from my house. She was there, packed the kids up and headed out to get me (AFTER asking if this was a joke -- but that was my own fault. For an April Fool's day prank a few years ago, I went out on a ride, then called her halfway out and told her I had an accident and broke my arm. Classic case of reaping what you sow.)
Kathy, a doctor, checked me out at the scene, then brought me home. I was pretty certain by now that I'd escaped the worst of it. She confirmed that my left side looked like a total mess, but I appeared to be otherwise uninjured. After that followed the most painful shower of my life; first, just the water hitting those raw wounds was horrific; then adding soap was another joyous activity. Kathy wrapped me up, and I spent a real fun afternoon and evening staying off my left side, adjusting bandages and having my family wait on me hand and foot.
This morning wasn't a walk in the park, either. I'm as sore as if I'd just finished a game at right guard for the Redskins. But all the parts seem to be working, no headaches and I shouldn't need a trip to the hospital. Overall, it could have been a lot, lot, LOT worse. I'll just need some time for the bruises/cuts/abrasions covering my side to heal up.
If I could pass along one piece of advice from this, it would be to shout ALWAYS, ALWAYS WEAR A HELMET! I'm quite sure it saved me from a serious head injury at the very least; at best, it might well have saved my life. I never ride without a helmet; it's fitted properly, and on tight. Since I had it properly secured, it didn't slide or come off on impact; it protected me exactly the way it's supposed to. I've done a lot of stupid things in my life; this might have been the smartest.
It's something all of you who ride, or who have children that ride, should remember. I can't tell you how often I see kids in my area, including very young kids in my neighborhood, riding their bikes without helmets. They and their parents, of course, never expect that they'll have an accident. I didn't expect to have one either.
And the cat? When I finally gathered enough of my wits to look around, the little mischief-maker was gone.
Posted by Keith Ward on 04/19/2008 at 12:48 PM11 comments
Long time, no blog. Sorry about that; I've been out of town a number of days on various job-related projects, and catching up on the work I missed while I was out on the other days. I wanted to tell you about one of those projects.
Yesterday I was in Baltimore, MD, at an elementary school that's using an extremely interesting virtualization technology. The school, Windsor Hills Elementary, has 200 students. Like most schools, they're struggling with funding. In order to get more computers in the classroom, they're using a device from nComputing that allows up to three computer stations to share the resources of one desktop. The device is a little black box, about the size of a modem, with a big "X" on top. The devices are hooked up, via standard ethernet, to a half-size PCI card added to the host desktop. The devices have a monitor, keyboard and mouse plugged into them. They are responsible for virtualizing the desktop image, pulling it from the host, and pushing it out to the workstations.
The school is using the X300, and in the classrooms I visited, there were two units hooked to a garden-variety IBM desktop with 512MB of RAM. Thus, three students were essentially sharing one computer. I interviewed a number of folks there, including the principal and a systems administrator; they both said the savings over buying a full desktop were substantial, and they couldn't have done this without the nComputing device. Instead of one or two computers per classroom the old way, they will now be able to double or triple that number in the future, giving more kids more time to use computers. Without this, that simply would not have happened.
Note also that these are standard desktop computers, not quad-core screaming servers or blade firebreathers. You don't need to upgrade hardware in general (although the school system is strongly considering doubling the RAM in the host desktops to help out), and don't need to add a single server in your datacenter to use the nComputing devices. Very, very cool.
From what I could tell, the students didn't notice any difference between working at the desktop computer and working at the nComputing workstation. They were playing games, surfing the Internet and doing schoolwork, and never knew what was going on in the background. In my opinion, this is groundbreaking technology, and something to get very excited about.
One more note: I was there with a colleague, and we were videotaping our visit. We'll be putting this together and releasing it soon on the Website as a story. Keep an eye out for it, as well as other video content.
Posted by Keith Ward on 04/17/2008 at 12:48 PM7 comments
First week on the job and I¡¦m truly excited to be joining Virtualization Review as Executive Editor, working to help fulfill the strong vision that Keith has for this publication. I came to VR from market researcher IDC, where I was involved, among other things, in looking at the business models and trajectories of emerging markets and the impact of IT hardware and software development on selected vertical markets.
One thing that has struck me about virtualization is the level of excitement around this new technology in all its manifestations. I¡¦ve been in the industry for a while and have seen many technologies come and go. But when measured against this perspective, the buzz around virtualization is pretty phenomenal. This is a technology and a market that has legs. I believe that tracking the trendlines that will drive this market through the twists and turns to come will be crucial to understanding its future direction. These trends will determine not only which vendors will have the best price/performance on an individual basis but also --- more importantly -- the business models and go-to-market strategies models that will shake out for applying virtualization across the spectrum of enterprise systems and solutions that exist today.
To help our rapidly growing readership get a better handle on these trends and developments, Keith has asked me to guide the development of our news and products section section, Upfront. In this section, we¡¦ll be looking at a host of different issues including but not limited to pricing models, the need for security, and the development of new management tools on the vendor side
Posted by Keith Ward on 04/08/2008 at 12:48 PM2 comments
VMware, apparently deciding that shorter is better, has truncated the names of its two hypervisors. The hypervisors formerly known as "ESX Server" and "ESX 3i" have been re-christened simply "ESX" and "ESXi".
No formal announcements of the name changes were made. An entry on the VMware Technology Network (VMTN) blog by VMware employee John Troyer contained the only company reference to the changes.
ESX is VMware's flagship hypervisor, while ESXi is the lightweight, embedded hypervisor that is available from major OEMs like Dell, HP, IBM and Fujistu.
It seems a little odd that VMware didn't make any public pronouncement about the name changes; it just kind of quietly did away with the names. No explanation was given.
Is this a good idea? I would have to say no at this point. The names are so close together now that there could be some serious product confusion. Although the idea behind both is similar, i.e. they're both hypervisors, their usage scenarios are very dissimilar. ESXi is much more limited than ESX in capabilities, and the codebase is substantially different, although future editions of ESX will be based on the code from ESXi. It will be very, very easy to mix up the two versions when ordering them, for example, or searching for them on the Internet or VMware's Website.
What do you think? I'd love hear from you.
Posted by Keith Ward on 04/07/2008 at 12:48 PM2 comments
It's a question increasingly asked these days: how does the Xen open source hypervisor compare with VMware's hypervisor, ESX Server, in terms of performance? One big reason it's asked is because of Xen's much lower upfront cost -- if it's in the same ballpark as ESX in terms of speed, that could have a big impact on a company's buying decision.
A preliminary answer is that Xen compares favorably in some ways, but in others it falls well short of ESX Server. That, at least, is the verdict of load testing done for the BriForum conference in Amsterdam last October. BriForum is run by Brian Madden, an independent virtualization and application delivery expert with a popular Website.
In the session, available as a video Webcast, Ron Oglesby of GlassHouse explains how Xen Enterprise fared against ESX Server in several different tests on a 1U Dell server. The first was a single VM running Microsoft Terminal Services. The two hypervisors compared well from a performance standpoint in terms of processor and memory utilization; in other words, the hypervisors were about equally efficient.
Things changed, however, when three more VMs, including a file server, Web server and domain controller, were added. The load testing was done by the open source I/O measuring tool iometer. At that point, VMware pulled away dramatically. With multiple workloads, ESX had almost three times the throughput and three times the disk I/O performance as Xen. That substantially better I/O performance led to much quicker response times for ESX, in both reads and writes.
In this analysis, it looks like Xen is a good candidate for lighter virtualization workloads, and VMware is better for heavier usage. Now, that being said, it's worth noting that since Citrix bought XenSource last year, the company has put an awful lot of resources behind developing XenServer (formerly XenSource), including new products for management and application delivery that improve its enterprise worthiness. That includes the just-released XenServer 4.1, with a new per-server price structure that throws down the gauntlet to VMware and its per-socket pricing. XenServer 4.1 has a number of strong upgrades that certainly make it worth a look.
Also note that XenServer is a Citrix-specific product; lots of companies use Xen as their hypervisor and build functionality on top of it, including Novell, Virtual Iron, Sun, Red Hat, Oracle and others. Like Citrix, they aren't selling the hypervisor; rather, they push the management and efficiency of products built on top of it.
In any event, the video is well worth a look, whether or not you agree with the results and analysis.
Posted by Keith Ward on 04/04/2008 at 12:48 PM7 comments
Interesting
tidbit from Alessandro Perilli over at Virtualization.info about VMware executives selling some shares. The list included President and CEO Diane Greene who, according to the report, sold some at the end of January 2008, and Carl Eschenback, Executive Vice President of Worldwide Field Operations, who did the same thing in early March.
Perilli speculates that it might have something to do with the decline in VMware's stock since its historic IPO last summer. It's gone even lower since then, and is now below the inital post-IPO glow.
One posting under Perilli's entry, from some who identifies himself as Ramesh Dharan, says it's no big deal:
"Practically all senior executives at major companies set up fixed automated share sale schedules, the timing of which is explicitly out of their hands. They do this to indemnify themselves from accusations of insider trading."
Another poster agrees with that opinion, while a third argues that any decline in stock is more due to the performance of EMC, VMware's parent company. While Dharan may have point, it doesn't mean that the execs in this case were just following an automated schedule (conversely, that could be exactly what happened.)
VMware is still a vibrant, healthy company in the vanguard of the virtualization movement. It has more competition than ever, which is certainly a good thing. But as of yet, I certainly see no cracks in the foundation.
What do you think? Is VMware on the downward slope, or still moving up (not their stock, but the company)? Tell me!
Posted by Keith Ward on 04/03/2008 at 12:48 PM6 comments
I spent a portion of my morning learing about Microsoft Virtual PC 2007, and found it to be...shall we say...lacking.
Certainly, it's a good value: As Homer Simpson would say, "No payments per month? Yeah, I think we can swing that!" Here's the background: the computer my kids use has Vista as the OS. There are a number of programs they use that haven't been updated for use with Vista (shocking, I know), so I wanted to load XP in a virtual machine (VM) and have them use the programs (mostly simple games) in the XP VM.
Downloading the bits didn't take too long. Then the typicall install process, which was pain-free. I created a VM with about 1GB RAM and a 30 GB virtual hard drive (.vhd, where the files are kept), and loaded XP (with SP2) on it. So far, so good.
The fun really began when I tried to load the first game into XP. It's called "Jewell Quest 2", and worked fine in an older XP machine that has since been decommissioned. It loaded the game OK, but when we tried to play it, the animated cursor the game uses went haywire. It was nearly impossible to control in the VM.
Thinking it might just be that game, I shut down the VM and restarted it. When XP had fired up, I loaded another game (Nancy Drew -- yes, that Nancy Drew, who has now become a series of video games) into the DVD tray, expecting it to find the DVD the way it did previously.
No dice. Nothing came up except an error message: "Unable to open CAL or CIF TREE file...Check path in your .ini file". That was it. I tried rebooting the VM again. Same message.
So far, Virtual PC is about as successful with these games as Michael Jackson is at looking like an earthling. Later in the day, my daughter did manage to get one game working correctly, so that lifts Virtual PCs batting average to .333. Good for a major-leage ballplayer, not so good for a VM.
Note, of course, that your mileage may vary. This is a very, very small sample in one household. However, if I'm having problems, given that I'm a power user and editor of a virtualization magazine, how many problems are other people, who don't play with this stuff all day, having? And how quickly will they give up on virtualization if it doesn't deliver on the main advantage of PC virtualization, which is using an alternate OS to run programs incompatible with the host OS?
Peter Varhol, who did a product roundup in our launch issue that compared offerings from Microsoft, VMware and Parallels, had good things to say about Virtual PC, so it may be just me. But so far, I don't think I'll be turning to it very often. Beyond the advantage (and a potentially significant one) of being free, I don't know, at this point, why I would choose it over competing products.
Posted by Keith Ward on 04/03/2008 at 12:48 PM6 comments
If you don't know
Andrew Kutz, you need to. He's a developer who writes (among other things) utilities for VMware. I haven't personally used any, but many in the industry swear by them. He's one of the smartest folks in the industry, and understands virtualization in a way few others do.
Kutz is also a prolific author who blogs frequently about virtualization topics, with a special emphasis on VMware. He's writing an article for Virtualization Review magazine for our next (May/June) issue about embedded hypervisors vs. packaged ones -- specifically ESX Server vs. ESX 3i. He definitely isn't afraid to say what's on his mind, and he told me that he "has a lot to say about embedded hypervisors." Should be very interesting.
Posted by Keith Ward on 04/02/2008 at 12:48 PM4 comments
My
blog item yesterday quoting Citrix's Simon Crosby brought a quick response from a VMware employee. The e-mailer wanted to remain anonymous, but I can tell you a few things about him: He's been with VMware for more than three years and "has extensive experience in large scale data centers running production VMware environments including many from the Fortune 100." Note: I have not edited his comments; they appear as he sent them to me.
First, he addresses the new licensing model for XenServer and related products, which is now per-server rather than per-socket.
"What I find ironic about the Citrix marketing and Simon's constant diatribes, is that their products are not backed by any significant adoption and thus are constantly using price as a way to make their point. If price is their only competitive edge, there is little room for growth. Low end will not carry Citrix/Xen to the next level in my humble opinion. They must innovate. A business model which is dictated by bargains and giving their product away is surely not to succeed."
He also takes aim at Crosby's assertion that ESX 3i, VMware's embedded hypervisor, is a "disaster".
"Virtualization customers know where to put their money. The fact about Citrix's product is that while it is certainly up and coming, they will always be two steps behind. It's interesting to watch them criticize VMware's direction yet copy our solutions the very next day. 3i is a perfect example. It is not a disaster at all, to say it is a disaster is to claim VI3 itself is a disaster as it's the same code base without a console OS and it's inherited problems. They do not want to admit that where VMware is going is the correct path because the[y] are unprepared to detach from the operating system."
My e-mailer also takes on questions about the price of VMware solutions, and says Crosby should choose his friends carefully.
"Simon should spend his effort improving his product and allying himself with VMware. Because of his lust for hating VMware, he ended up partnering with Microsoft who is a direct competitor for SMB's which is the only hope for both MS and XenSource. In effect, they are partnering to compete directly for the same market, this is very strange. Meanwhile virtualization customers are laughing at this because those who are "in the know" do not care about price because VI3 pays for itself upon implementation anyway at the very minimum. In the end, the minor cost savings are irrelevant compared to the advantages. One can argue that VMware is cheaper besides all of the marketing going on by the competition. You get a lot for your money with VMware, and have the option of spending very little as well. The pricing scales quite well. Consolidation rates with VMware far exceed the competition anyway."
He adds this final parting shot at Crosby:
"I wish Simon luck. He's going to need it if he continues to step on VMware who his target market currently adores."
The issue of VMware pricing is very significant in the industry today. There is no doubt that the upfront costs for ESX Server and VI3, the infrastructure product, are much higher than the competition -- note that you can download the Xen open source hypervisor absolutely free, and Microsoft's Hyper-V is just $28 for a stand-alone version (and free with 64-bit Windows Server 2008). But VMware, and some customers I've talked to, believe it's a good investment that ends up saving money.
Of course, users of XenServer and other competitors would make the same claims, with a much smaller upfront investment. The question, then, becomes if VMware products offer such superior advantages that it ultimately ends up being a better value. I'd love to hear from those of you who are using VMware and another solution like Virtual Iron, Microsoft, Virtuozzo from Parallels, or some flavor of Xen. What do you like and why?
Posted by Keith Ward on 04/01/2008 at 12:48 PM4 comments
Hiya, gang! I wanted to let you know that we're starting to put new features on the Web site now, including content from the launch issue of the magazine. One of the most exciting features is the debut of our Virtual Advisor columnist, Chris Wolf. You can find his premier column
here. It's about virtualization licensing -- which companies are doing it right, and other companies that are, shall we say, behind the learning curve a bit.
The print edition of Virtualization Review should be hitting your mailbox any time now, assuming you're a subscriber. I'm very proud of the work the team has done, and think you'll be impressed with it. Feedback is more than welcome; it's required if we're going to make this magazine -- and Web site -- the best it can be.
Posted by Keith Ward on 04/01/2008 at 12:48 PM1 comments
VMware's Lifecycle Manager is officially
available. It's basically a lifecycle and chargeback product, allowing admins to trace a virtual machine's (VM's) creator, owner and death date (i.e. when it's scheduled to be decomissioned. Aren't you glad humans don't come with similar information?)
Having chargeback ability is nice; with VMs so easy to create, it's equally easy to lose track of which department is using what, making chargeback a real business challenge. Since I'm not a big fan of rewriting press releases, I'll quote at length VMware's release about some of Lifecycle Manager's core functions:
- "Create a catalog of standard IT services. Users can select from a pre-defined menu of virtual machines with different properties such as processor and memory. This standardization allows infrastructure administrators to maintain control over the IT environment and minimize risk.
- Streamline requests and approvals. VMware Lifecycle Manager establishes a consistent and scalable mechanism to route and approve all requests for virtual machines, ensuring compliance with internal policies.
- Track and control virtual machines. VMware Lifecycle Manager provides an easy to use Web interface for tracking virtual machine deployments so IT administrators can know exactly when requests were made, approved or denied; when and where virtual machines are deployed; and how long they have been in operation.
- Eliminate manual, repetitive, and error-prone tasks. As virtual environments grow, automation is critical in helping IT do more with less. VMware Lifecycle Manager automates each step in the virtual machine lifecycle based on predefined policies.
- Assign Chargeback Metrics. VMware Lifecycle Manager enables IT to associate chargeback metrics to specific virtual machine deployments and resource pools. These chargeback metrics can be assigned to specific business groups, and tied in to existing financial systems.
- Integrate with existing management tools. VMware Lifecycle Manager provides APIs for integration with other IT operational tools such as trouble ticketing, change management, and asset management."
There are many tools available from third-party virtualization vendors that do the same thing; one attractive part of VMware's offering is that it bundles everything into one package.
Is this something you've used in beta, or does it intrigue you? Let me know what you think of Lifecycle Manager.
Posted by Keith Ward on 04/01/2008 at 12:48 PM2 comments
As a technology journalist, I'm used to marketingspeak: you know, the kind of language you find in press releases -- "Our new product is a game-changer!" Or, "This is a big win for customers!" Or, "Our company, the leader in _____ (fill in with any technology)." It sometimes gets wearying, which is why I enjoy talking to Simon Crosby of Citrix.
Simon says what's on his mind, doesn't pull punches and isn't afraid of anyone -- least of all VMware, the market leader in virtualization. For example, when talking about the new embedded hypervisor for HP servers -- HP Select Edition -- Crosby compares it VMware's ESX 3i, which is also an embedded hypervisor. "ESX 3i is just a disaster, and we love it," Crosby told me in an interview last week.
Crosby, many of you will know, is the former CTO of XenSource and the CTO of the Virtualization and Management Division at Citrix Systems. As such, he comes from an open source background, and has a lot of that maverick mentality that you tend to find from the open-sourcers.
Open source projects frequently compete against well-established, entrenched companies with huge R&D budgets, and hence have that spirited David vs. Goliath outlook. Although Crosby now works for a billlion-dollar corporation, he doesn't seem to have lost that sharp edge. Here are some more juicy quotes in his war of words against VMware and his take on HP Select, which is an HP-tuned, embedded version of XenServer. "We have better product, they [VMware] charge for [theirs], and we're cheaper. I would love to see more products like this from VMware." In other words, he'd like to see VMware produce more products that he thinks are trounced by competing Citrix offerings. Now, is that fun, or what? Note that I am not evaluating the truth of his statements; I'm just saying that it's a breath of fresh air in a world of cloying P.R. language.
One of Crosby's central theses in his critique of Citrix's virtualization strategy vs. VMware is disruption. What he means is that in his view, VMware is a much more disruptive technology in that it requires wholesale changes to an I.T. infrastructure, whereas Citrix works with what you've got. "VMware wants to own [the virtual infrastructure]," Crosby says. "We want to be part of it. The right thing to do for [virtualization] adoption is to make virtualization subservient to existing systems."
That may or may not be true -- lots of companies are having success implementing a VMware infrastructure. In fact, one flooring company IT admin -- Kevin Sonney of iFloor -- told me last week he's consolidating everything on VMware, and it's saved him many thousands of dollars, allowing him to do things he never could have done without VMware. He said that instead of buying the 20 servers he would have had to in years past, he bought four, and consolidated everything onto those boxes. Those are real, tangible savings through virtualization.
As you can tell, Sonney is happy as could be with his VMware infrastructure, and he's far from alone. But Crosby also threw out another very interesting tidbit: pricing for XenServer is now per-server, as opposed to per-socket. With multi-processor servers and blade servers fast becoming the norm for virtualization environments, that's a compelling and aggressive pricing move. And yet another salvo in the pricing wars. It will be interesting to see how -- and if -- VMware responds. VMware, like the majority of vendors, holds fast to a per-socket licensing model.
What's your take on Crosby's thesis that it's better to work with what you have, rather than overhaul your infrastructure, when it comes to virtualization? Let me know.
Posted by Keith Ward on 03/31/2008 at 12:48 PM3 comments