I just completed some consulting work for my former employer, Novell, and had the opportunity to work once again with Sal Darji and Ed Murphy. During the course of this work, I again ran into something that has long baffled me. Why is anyone buying proprietary Unix servers anymore?
To see why I say this, perform a simple comparison for yourself. Take the SPEC server CPU benchmark for a proprietary Unix server from Sun, IBM or HP, and divide it by the list price for the server. Do the same for a commodity Dell x86/64 server. Compare.
In my analyses, I use SPEC’s SPECint_rate2006 benchmark. It measures integer processing, which is more typical of business workloads. Also, this benchmark measures throughput rather than peak speed, since most datacenters are more concerned with the number of transactions processed, not minimizing the time for a single transaction. (The peak speed benchmark, SPECint2006, would be more appropriate for realtime computing.)
The value of the SPECint_rate2006 benchmark for a given server represents its capacity. We can treat SPECint_rate2006 as a unit of capacity, and when we purchase a given server, we are adding capacity to our datacenter equal to the SPECint_rate2006 benchmark for that server. Given this, the cost per SPECint_rate2006 is the amount we’re paying for each unit of additional capacity.
I realize that this analysis only holds true if all we really care about is throughput, and integer processing. Every workload is different, and the results may be different if the characteristics of a particular workload differ significantly from this approximation of a typical workload. But still, the results are startling.
Based on publicly available list prices, Solaris servers range from $300/SPECint_rate2006 to over $1,000. In comparison, Dell x86/64 servers range from $30/SPECint_rate2006 to no higher than $170. At the low end, this is an order of magnitude difference in cost/performance! And it’s not just Dell — IBM, HP and Sun x86 servers have significant cost/performance advantages over their own Unix servers (although not as large).
This cost advantage could get eaten up by operating system licensing costs. If Windows were the only alternative on x86 architectures, that would certainly be true. But not only is Linux cheaper than Windows, it is typically as cheap or cheaper than the license costs of the various Unixes.
At the high end, Unix servers are much larger than the largest available x86 servers. But the Unix world is moving towards virtualization, with single servers running multiple workloads. In this scenario, there is no advantage to a scale-up architecture vs. a scale-out architecture. Besides which, with quad-core x86 chips and ever more processors available on high-end x86 servers, the gap is steadily shrinking. I can see that occasionally x86 servers may not be able to meet the HPC needs of a particular app, but this would have to be the rare exception rather than the rule.
I have heard anecdotally of significant discounts off list for Unix servers, much more so than for Dell servers, and it’s easy to see why. Perhaps the Unix vendors are able to discount low enough when they have to so that the switching cost eats up any initial cost savings. And of course application and middleware software compatibility can be an issue, but these days almost any vendor offering software for Unix also supports one distribution or another of Linux.
So I remain baffled. Given these economics, for the vast majority of workloads, why would anyone stick with Unix? I realize the market has been steadily moving away from Unix and towards Linux and Windows, but plenty of shops are still buying new Unix servers for brand new applications. Why?
I am entirely open to the idea that I’ve overlooked something, so if anyone has any thoughts about additional factors to consider please leave them in comments. But until proven otherwise, my recommendation to any datacenter is to avoid Unix except for the rare exception, and move to Linux on x86/64 instead.
I have been doing some contract work this month with old friends and colleagues now at Telogis. The Novell and Cambridge connections are legion — Dave Cozzens, my old boss and manager of the Cambridge Digital Business Strategy practice is now Telogis CEO, and many former colleagues are either working there or contracting. Even Jack Messman, the former Cambridge and then Novell CEO is on Telogis’ Board.
Just like homecoming for me.
Telogis has a fleet tracking SaaS application, OnTrack. I’ve known about fleet management and tracking systems going back to the old satellite days and before, but I was surprised at how far things have come. Telogis’ technology reflects an accumulation of incremental changes, none of which is horribly earth-shattering by itself, but cumulatively is pretty interesting.
It turns out that heavy vehicles, like all consumer vehicles these days, have onboard computers, and there is an industry standard for querying those computers via a protocol called JBus. Telogis has onboard hardware with GPS and a wireless modem. Add a standard piece of hardware to query the vehicle computer and pass it to the Telogis hardware, and you can get close-to-real-time reporting of everything the onboard computer knows, plus GPS. Telogis can effectively re-create the vehicle dashboard, including not just speed and location, but things like battery voltage, oil pressure, engine temp, even whether the seatbelts are buckled and the doors are latched.
Fleet tracking started off as a way to monitor drivers, to make sure they weren’t parking somewhere for a nap or using the trucks for side jobs, and monitoring drivers is still part of the value. Now employers know if their drivers are speeding and what routes they are taking between destinations. But the value is expanding to monitoring the vehicles themselves — to knowing ahead of time when a vehicle needs maintenance, say because the fuel economy is declining or the engine is running hot. Add sophisticated routing algorithms for dispatch on top of this, and the vehicle, driver, customer and corporate office are all highly integrated.
I know this is all rather mundane, and familiar to anyone working in a company with field service or delivery operations. But it’s an example of what can be done with location-based computing and wireless networks. This is one of those things that has quietly advanced without much fanfare, and turns out to be rather nifty.
This past Saturday morning I attended a conference at Cal Tech here in Pasadena on web platforms. The similarities between operating system platforms and web-based platforms really struck me. Specifically, it seems there is an analog to the FSF’s four freedoms of FOSS and the freedoms provided by various web platforms.
These freedoms are not to be confused with Marc Andreessen’s ontology of web platforms. His typology is architectural — more akin to the distinctions between operating systems, middleware and applications. But among Andreessen’s Level 3 “run-time environment” web platforms (current and planned) such as Salesforce.com, Ning, Facebook, Amazon AWS and Google App Engine, there are different freedoms afforded to users.
- The freedom to use your own data. (See the social web users’ Bill of Rights.)
- The freedom to design a unique user interface (without requiring platform logos or “badgeware”).
- The freedom to develop your own apps (even if they compete with the platform’s own apps).
- The freedom to embed apps running on other (competing) platforms.
- The freedom to easily take your apps elsewhere.
This list certainly needs to be refined. But it’s a useful starting point to compare the various platforms on the market. Facebook has mightily resisted providing freedom number 1, resulting in a potential opening for Ning and others. Salesforce.com (their platform offering is called Force.com) promises freedoms 1-3, but to my knowledge doesn’t support freedom 4 and definitely does not support freedom 5. It isn’t enough to expose APIs if those APIs are custom and proprietary to the platform, since this locks in the user. Hence the impetus for Open Social, an initiative for standard platform APIs that wants to provide portability of applications.
For a user of the platforms, the more freedom the better. But this is where the analogy with open source comes in — use of the platforms must be monetized. In the consumer space this happens with ads, and in the business space this happens with by-the-sip pricing, but these freedoms can create opportunities for free riders, just as in open source.
But, more so than with open source, we need to remember that it’s very early days for web platforms. We’re seeing the concept of platform moving from an operating system to the entire web itself. It will take some time for the economics, the technology and the marketplace to sort itself out.
There’s the Suse gecko, the Mono monkey, the Higgins mouse, the Bandit dog, the Open Office seagull, and of course the Linux penguin.
Now there’s the fossa.
Novell has named our new architecture the Fossa Project because of the fossa’s renowned agility. (Ed Murphy, who has two younger children, informed me that the fossas are the bad guys in the movie Madagascar. Since my kids are older, I’ve never seen Madagascar, but I’m thinking they must be like the hyenas in The Lion King. Kinda takes all the fun out of our new mascot.)
The Fossa vision (Novell’s, not the animal’s) does a good job of bringing together Linux, virtualization, data center orchestration, identity (both for users, devices and apps) and collaboration. For Jeff Jaffe’s presentation on Fossa, go here (Real video). For documentation on Fossa, go here and click on the Presentations tab.
Updated below. And as a reminder, this is my personal blog, and I don’t pretend to represent the views of my employer, Novell.
With Oracle acquiring Bridgestream and now Sun acquiring Vaau, I know all of you are asking yourselves “I wonder what Bob thinks about this?”.
No? Well, I’m going to tell you anyway.
Novell decided earlier this year to develop our own roles module, which you might think killed any chances of us partnering with third party RBAC vendors, but not so. Just last month we conducted a joint project with Eurekify, a relationship that both sides are looking to expand. So why would we develop a roles module in-house instead of acquiring an existing vendor? And once down this path, why still partner with Eurekify?
There are two distinct pieces to RBAC software. First is the software that computes user privileges based on roles and provisions them. This functionality requires a very specialized database that is highly tuned for performance in a rather narrow set of tasks dealing with identities and entitlements. We call this specialized database a directory, and Novell happens to have one. For role-based provisioning to work effectively, it needs to be built on a directory. Acquiring a third party product and integrating it into an existing directory requires a massive re-write. So we didn’t do it. (Actually we did acquire some third party RBAC software that was already built on top of eDirectory, and have been enhancing from there.)
The second piece of RBAC software is the roles mining and analysis that supports the development of optimal role definitions. This functionality requires a lot of ad hoc queries and computations based on matching many different pieces of data. This workload has a very different profile from the straight role-based provisioning, and requires the flexibility of a relational database.
Vaau, Bridgestream and Novell’s roles module all support the first category. Vaau and Bridgestream are trying to also compete in this second category, but they are playing catch-up. The clear leader in role mining and analysis is Eurekify.
Vaau and Bridgestream are trying to be all things to all people, but they will face the challenge of, on the one hand, integrating with Sun’s/Oracle’s directories to support role-based provisioning, and on the other hand, trying to expand the analytical capabilities that can only be built on top of a relational database. They’ll be busy for awhile.
Meanwhile, Novell is not trying to pretend we compete in the role mining and analysis space. We are providing a highly tuned role-based provisioning capability based on eDirectory, and we will partner with Eurekify for the things a directory, and a directory company, can’t do well.
(Now, to kill any rumors, I know of no acquisition talks with Eurekify, or the lack thereof. I’m not being coy. I truly have absolutely no information that is not already in the public domain. Actually, I have less, since I’m sure there’s a lot that is public that I haven’t seen.)
So the Bridgestream and Vaau acquisitions are somewhat of a “me-too” strategy, matching Novell’s direction of integrating role-based provisioning into our IdM suite. But the acquisitions carry significant added challenges as well. And they preempt Oracle’s and Sun’s ability to partner with Eurekify for role mining and analysis, which given Eurekify’s leadership in this space, they may come to regret.
Update: Novell has announced our roles module, named the Roles Based Provisioning Module.
I suppose I should say something about all the fires in Southern California, given how much of my attention it’s been occupying lately. I spent the three weeks before the fires at a client down in Rancho Bernardo. I wasn’t scheduled to be there this week, but needed to talk to the Novell team still down there first thing Monday morning. I heard about the fires, but didn’t realize they had hit RB until I called a member of the team on his cell and caught him at the San Diego airport on his way out. RB evacuated, project scrubbed for the week, our whole team going home.
My younger brother and his wife live in RB — I called them immediately, and learned they too had evacuated and were staying at a friend’s in Poway. (Yes, parts of Poway were later evacuated, but luckily not my brother’s friend.)
I was born and raised in San Diego, and I know the neighborhood where all the houses in RB were lost. One apartment complex that was partially destroyed, La Terraza, is right across the street from the house my parents lived in for years after I left home for college. It’s all boringly middle-class, definitely not the upper-income-houses-nestled-in-the-hills stereotype of homes that get hit by fires. Nearby Ramona, hit the hardest, is even less affluent, largely rural kind of place that’s been there for years, so don’t listen to those people who want to blame the victims for living in areas at risk for fires.
If you want to find a political angle to this, consider that San Diego is a rabidly anti-tax county. Voters have rejected ballot measures in the past to create a county fire department and to spend more on fire protection. Perhaps that attitude will change.
So, my brother’s place is fine, and so are my client’s offices. Life will start to return to normal next week. But hopefully San Diegans won’t forget too quickly.
My employer, Novell, posted its earnings yesterday, resulting in a nice up-tick in our stock price. Press and commentary has been favorable for the most part, but I feel compelled to respond to Matt Asay’s two posts on our earnings.
Now, I am an employee of Novell, so to some extent my fortunes rise and fall with those of my employer (although there are always other options in a free labor market.) But I am not in PR, and I try to make my comments on this blog objective and fairly independent. I do self-censor along the lines of “if you can’t say something nice about your employer, don’t say anything at all”, but I’d like to believe I’m able to see all sides of Novell’s position relative to competitors, peers and partners.
But I’m afraid to say that Matt Asay, a one-time Novell employee, has lost the ability to view Novell through any lens other than his distaste of the Microsoft deal. Novell had a pretty nice quarter, with our stock up over 8% today as I write, but this is what Matt says regarding our Linux business:
The numbers are still relatively small ($21 million), but any progress is progress. Importantly, the word on the street is that Microsoft is not proving to be a great distribution partner (go figure!). This means that (gasp!) Novell is actually managing to sell its Linux value, not just making silly patent deals.
Sure, the patent deal between Microsoft and Novell was a deal winner at Wal-Mart Stores and undoubtedly a few others. But I would argue that Microsoft has done more harm than good to Novell’s Linux business. Why not cut it adrift?
Matt would argue that the MS deal has done more harm than good, but he doesn’t — he just asserts that this is the case, while linking to a nine-month old article. The actual evidence provides a different picture: as Matt himself quotes from Credit Suisse, $14M in Linux revenues were from the MS deal, while $24M were “organic”. “Word on the street” may be that Microsoft isn’t a good partner, but the quarterly results tell a different story altogether.
Matt excerpts Jason Maynard’s take on the earnings report, which says in part:
Although we acknowledge the improvements to the business, we continue to believe there are better relative investments within the group….we are maintaining our Underperform rating….
So CSFB has been pretty bearish on the stock, and they’re not willing to change their rating after one or two good quarters. Fair enough. Maynard makes various arguments for his view, some of which I think are valid, and some off-base. Of course Maynard is just one analyst, and others have been more bullish, but none of them make suggestions as far out as Matt does:
Now you just need to ditch the ballast that you call Systems and Resource Management and Identity and Security Management, and really grow…
Interesting…so Matt advises getting rid of almost $100M in quarterly revenue and about $35 in quarterly operating income, as well as divesting the customers that own SRM and ISM products but not Linux, because…why?
In most businesses, 2 percent and 4 percent increases are embarrassments.
Matt’s background in hi-tech start-ups is showing. Start-ups can’t survive without rapid growth, but the rest of the business world is hardly embarrassed by a slow growing but profitable business. While start-ups can’t worry about profits at the expense of growth, for the rest of us, profits are a good thing. Novell has a stated goal of growing these business more rapidly, which would be good, but in the meantime they’re not exactly a millstone around our neck.
I’ve heard analysts argue that Novell should shed everything but Linux so that it will have the same business model as Red Hat, a view Matt seems to be alluding to. Sure, Novell could remake itself to look just like Red Hat, but then Novell would throw away most of its revenue and income as well as any chance it has to differentiate itself from Red Hat. The world doesn’t need another Red Hat, and Novell doesn’t need a “me-too” strategy. It needs to find ways to deliver more value in ways that Red Hat can’t. One way to do this is with a broader, integrated product portfolio. I think the chances are better that Red Hat will look more like Novell down the road rather than the reverse.
Matt finishes with the comment “Novell needs a new trick”, by which I assume he means something besides our deal with Microsoft, even though only $14M of our Linux revenue was from Microsoft certificates, leaving $229M in revenue from other “tricks”. We’ve announced a string of partnership deals, an acquisition, and several new product releases in the last quarter, but Matt overlooks all of that due to his over-riding obsession with the MS deal. How many more tricks does he want?
Matt and other open source commentators should feel free to criticize Novell’s MS deal on its merits, or lack thereof. I think they are wrong, though, to view all of Novell through that single lens. They also are wrong to think that Novell, a complex multi-business unit company, must look like Red Hat to succeed.
None of us at Novell know how this wild ride is going to turn out. We may succeed fabulously, or we may crash and burn. Either way, I’d like to see a higher quality of kibbitzing from industry observers than Matt has displayed.
IBM and Sun have announced an agreement for IBM to resell Solaris on its x Series servers. Mark Webbink believes (and Matt Asay concurs) that this is a Machiavellian move on IBM’s part to keep other OS vendors on the defensive:
For IBM it provides another Unix-based operating system vendor with which to challenge Microsoft and with which to keep pressure on the other Unix-based operating system vendors, Red Hat and Novell.
Mark Webbink has been Red Hat’s top lawyer, so he clearly has had far more exposure to the top dogs in the software industry than I have, but I have a much simpler explanation for this move.
IBM wants to sell hardware.
I mean, jeez, we can read all kinds of portents from the entrails, but I’m with William of Ockham on this one. Sun is getting out of the hardware business, they have ported Solaris to the x86 platform, they have a very large user base with servers coming off warranty, their customers don’t want to have to port their applications, and IBM sells x86 servers. What could be simpler?
From the slashdotted article on the news:
Left out of the mix here is Hewlett-Packard, which is locked in a battle with IBM for leadership in the worldwide server market. IBM and HP each had 29 percent share in the most recent figures compiled by market tracker IDC, while Sun and Dell Inc. were tied for third with 11 percent each.
That’s the real news here. This raises lots of interesting questions — who has the upper hand competitively here? Did IBM get an exclusive, or are they only getting first-mover advantage? Could Sun say no to IBM’s demand for an exclusive reseller agreement? Could IBM afford to say no if Sun refused an exclusive? Did the negotiations come down to a term for an exclusivity deal? Was Sun shuttling back and forth between HP and IBM to get a better deal, or are there strategic reasons why they preferred IBM? And what about Dell? Not enough Unix experience?
These are the interesting aspects to this announcement, not its impact on Microsoft, Red Hat and Novell, in my arm’s length view. As much as we’d like to think the OS wars are the center of the universe, others have their own businesses to run.
So I go on vacation, and look what happens — Novell owns the Unix IP after all, leaving SCO a dead man walking.
There is quite a bit of commentary out there, to which I will merely add this thought: this is one more example of a severely broken patent system. As I’ve said before, I don’t buy in to the FSF’s free software ideology for a second, but I do not think software requires patent protection for a very pragmatic reason: software patents are harming innovation, and the only rationale for patents is to spur innovation. There are no fundamental property rights in question here, just as there is no fundamental right to software freedom. This is a question of balancing society’s ability to take and build upon others’ inventions versus the desire to provide innovators financial rewards to incentivize invention. Nothing more.
And with regards to software, this balance is clearly out of whack, as shown by the lawsuit masquerading as a public company that was SCO. For quite a few years, SCO’s only viable business has been its extortion racket based on spurious legal claims. Of course this extortion scheme has impeded others’ ability to create business value, thereby subverting the whole purpose of the patent system. And while we all knew how this would ultimately end up, it’s taken many years already, and probably will drag out for several more.
This is a case study in why software patent protection should be eliminated.
I had to laugh…I took a look at Microsoft’s new open source web page, and decided to tag it for future reference. When the del.icio.us window came up, there was an interesting tag on its list of both recommended and popular tags.
Apparently del.icio.us users see quite a bit of “humor” here (and understandably so).
(Click on the image to expand.)
« Older Entries