February Update
Feb 17It’s been a very busy start to 2011 with the new Lithium iOS app nearly ready for submission to the app store, updates to Lithium 5.0, preparing to distribute Lithium Core as a Linux-based VM Appliance and progressing Xsnmp from beta to production.
We have posted the Lithium Roadmap for 2011 online. It details what we have planned for this year for both Lithium 5.0 and introduces our most ambitious project yet: Lithium 6.
There’s a first look at the new UI in Lithium for iOS available on lithium5.com.
Xsnmp is progressing well with thorough beta testing and addition of RAID card monitoring via SNMP.
And finally, there’s a few bits and pieces on Scotch and Code.
Prediction: Active Storage’s Drop-In Xsan MDC Replacement
Jan 30At 6AM PST, Active Storage is set to make a big announcement that they’ve billed as “When One Door Closes, Another Opens“. Active Storage is known for their excellent RAID Enclosures that, in competition with the Promise VTrak units are the successors of another of Apple’s hardware cancellations — the Xserve RAID.
With the Apple Xserve cancellation announced on November 5th 2010, effective as of January 31st, the timing of this announcement let alone the form factor of the silouette on activestorage.com points to some sort of Xserve replacement.
Here’s what I think it is:
I think we’re looking at a linux based box running Quantum’s StorNext — a box that you can drop in as a replacement for your Xsan MDC with minimal change to your Xsan clients and underlying storage arrays and fibre channel switches.
Active Storage is a storage company who have found an excellent niche in the Apple space with a high-performance Xsan compatible RAID enclosure. They were, in essence born out of the need for a drop-in replacement for the Xserve RAID. With the Xserve gone, it makes sense that they would feel the need to fill that void too. Otherwise as customers look to StorNext as an Xsan alternative they may look at other storage vendors too.
Offering an off-the-shelf, ready-to-roll Xsan MDC replacement, running on Linux and StorNext would be a very sound strategic move to not only keep their existing customers but vertically extend their presence in the customers SAN without the need for major change in the customers existing SAN environment. It also neatly ensures that Active Storage is a storage vendor that’s not just tied to Xsan — a product that may or may not have a long future ahead of it.
On Xserve Replacements…
Will the box be a complete “Xserve replacement” complete with some sort of native or virtualized Mac OS X Server? I don’t think so. As per my previous posts on Mac OS X Server, I do not think we’re going to see a change in Apple’s licensing stance. They want Mac OS X Server clearly and neatly defined as being suitable for the scale of use that would suit a Mac Mini.
Perhaps though it will offer some like-for-like server functionality. After all, the majority of the services provided by Mac OS X Server leverage freely available, open source projects that are widely used in Linux and other UNIX operating systems.
Taking it further…
Here’s one intriguing possibility… It would be possible to reverse engineer the servermgrd protocol to allow a third-party server to be configured using Server Admin.app just like a Mac OS X Server host, even if that third party server wasn’t using Mac OS X Server. We’ve done extensive work in reverse engineering servermgrd, and I can see how this would be possible.
The third-party server, running Linux for example, would simply need to present a servermgrd-like interface that Server Admin.app could connect to. The host would need to accept the configuration commands from Server Admin.app and translate that into config for the services running on the Linux box such as Bind for DNS, Postfix for SMTP, Dovecot for IMAP, Apache for HTTP/Web, etc.
As long as the third-party server presented configuration and state information in the format that Server Admin.app expects, and accepted the commands that Server Admin.app executes then the third-party server would look and feel just like a Mac OS X host through Server Admin.app.
The longevity of this solution would be predicated on two things: firstly how well Apple took to the vendor reverse engineering their servermgrd protocol and interface, and secondly the life of Mac OS X Server itself. If Mac OS X Server is retired, then there’ll be no need for Server Admin.app and the effort that’s gone into making the third-party implementation of servermgrd would be wasted. However, Mac OS X Server seems here to stay at least for the moment with the Mac Mini server form-factor.
Live Call-In 6AM PST…
Active Storage will be holding a live announcement call at 6AM PST, 31st January to unveil this new box. Call-in details can be found on the Active Storage site.
Xserve Cancellation Re-Defines Mac OS X Server
Nov 10Apple has announced that the Xserve hardware platform will be discontinued and no longer available for sale after January 31, 2011.
The message I’m taking out of this is “Mac OS X Server does not belong in high-power, high-density 1RU server installations. If your needs require that scale of computing, storage and form factor then look elsewhere.” The cancellation of the Xserve has re-defined Mac OS X Server’s place in the market.
The complimentary message to this is, “Mac OS X Server is great for the small and medium business/educational market who want a Mail, File, Calendar, etc server and here’s two options for deploying it — Small and Medium”.
There’s a reason that you often hear the terms “Small” and “Medium” combined to describe a single category of users or organization. They have similar needs, just on a different scale. When you move into the Large or Enterprise scale of user then typically the needs themselves change much more than just the scale of the solution.
The growing uncertainty about the future of the Mac and Mac OS X hasn’t concerned me much until now. For those not familiar with me, I’m James Wilson and I wrote Lithium, the Network, Server and Storage application suite for Mac OS X and iOS. Losing the only high-end server platform that the back-end of our flag-ship application runs on is a big deal to me. Like it or not though, it’s happened and there are definitely consequences for many different people and organizations.
I’ve seen many people complain that rack density needed to replace their existing Xserve deployment with like-for-like Mac Pro’s would necessitate a big increase in floor or rack space. Though that’s an interesting exercise in finding fault with the decision, I am not aware of anyone who is intending to replace all their Xserves with Mac Pros off the back of this announcement.
The form-factor argument misses the point of the cancellation. If your needs would have required you to buy an Xserve, then Apple no longer has a suitable product for you. If your requirements mean you need a 1RU, high-performance, truly server-grade device then you need to look elsewhere.
Similarly I’ve heard the opining that the Xserve is the first to go, then Xsan, then Mac OS X Server altogether (and the more paranoid amongst them believe the whole Mac platform is next). I don’t think that’s the case, at least not in the short term. The recent announcement of partnering with Unisys for enterprise support of Macs; the huge improvements made in Mac OS X Server from 10.5 to 10.6; the creation of a ‘Server’ style form-factor for the Mac Mini; etc, etc. All these things tell me that Apple does see a place in the market for a server solution built on Mac OS X.
Here’s what we know: Apple cancelled an unsuccessful product that was not selling well.
Here’s what I believe: Apple cancelled a product that was not fit for purpose in light of where they want to position Mac OS X Server.
That the Xserve was not selling well can be proven on paper. The reasons why it was not selling well are subject to conjecture. Some say because Apple didn’t market it well, didn’t push it through their sales channels, etc. The implication being that Apple could have made the Xserve a successful product if they wanted to. In fact I believe that this is true. It’s a great machine. Beautifully designed, powerful, flexible and a solid contender in it’s market with an incredible advantage over the competitors — it can run Mac OS X and they can’t! But if the discontinuation of the Xserve RAID didn’t tell you that Apple wasn’t interested in the data-centre, then surely this spells it out for you.
I put forward that the waining sales and eventually cancellation has nothing to do with malaise or a sales force that wasn’t capable of selling the product. To me, the cancellation of the Xserve is about defining where Mac OS X Server fits in the market. It’s not about cancelling an unsuccessful model just because it wasn’t selling well; rather it’s about removing a deployment option for Mac OS X Server that did not suit Apple.
I am confident that Mac OS X Server does have a future — but only in a small and medium variety. That said though, I’m not sure that Mac OS X Server has as long term or certain a future as the iPhone or iPad. I think we will continue to see the development of a Mac OS X Server operating system, but only until the point where Apple changes the way in which we use our Macs, iPhones and iPad such that we no longer have a need for a server.
Why then has Apple sought to re-define the purpose of Mac OS X Server and in essence limit the scale on which it can be used? Because creating and maintaining a Server OS and hardware platform that enters into the realm of high-density, large-scale data centre deployments is clearly not Apple’s thing.
Personally I believe that a lot of what goes into making a brilliant server platform, storage appliance or other data-centre-centric device is somewhat counter to Apple’s modus operandi. These boxes need to be rock solid, refined to the nth-degree over constant iteration not reinvention — aesthetics are of minor concern. To this end, consider taking the amount of time and money Apple invested in designing the aesthetics of the Xserve and put it into designing a machine of similar specifications without the aesthetics. I would argue that there is a high chance you will end up with a product that is more suitable for the data centre in terms of reliability, performance and price.
For the end-user of the Xserve, there’s a need to re-assess your needs. If you can achieve the same with Mac OS X Server on a Mac Pro or a Mac Mini then there’s the viable alternative. If not, then you’re going to need to ask some bigger questions that is most likely going to involve a change in operating system. Either way, I doubt you’re going to change your desktop, notebook, phone or table device of choice — right? Exactly.
Now for me as a software developer, the cancellation of the Xserve has a different implication. It has placed quite a significant cap on the high-end deployment possibilities for one of my applications in particular — Lithium. Without the Xserve, we’re left without a viable data-centre friendly option for our customers. This is not something we can or will ignore. We are now working on revising our roadmap for Lithium in light of the Xserve cancellation and my interpretation of what is means for Mac OS X Server as a host operating system for applications.
