Inside Apple's Leopard Server OS

With the upcoming release of Mac OS X Server 10.5, aka Leopard Server, the Mac IT world is thinking "What is going to be new?" Well, to be honest, everyone is. Apple's infamous closed-mouthed approach to major OS releases, while great for marketing purposes, isn't always so great for the IT world. However, Apple isn't a road map company, so if we want to get an idea of what to expect in Leopard, we have to dig into the public information Apple has released.

Fortunately, there's a decent bit of it. It's not everything, (certainly not Steve's famous "Top Secret" features), but it's not a sharp stick in the eye either. One of the big features for many people is the iCal server. I know I've not been alone in saying that the lack of calendaring in Mac OS X Server has long been a real hole in the product's feature set, and with iCal Server, Apple is making its first real attempt at plugging that.

iCal Server

iCal Server is an open source-based calendar and scheduling server with per-server licensing. So unlike Exchange and many other calendaring servers, you can run as many clients as you can against the server for the same price. (Regardless of how iCal Server stacks up feature-wise against Exchange and the others, the licensing model for iCal Server is one that works out much better for customers.) iCal Server allows for both human and resource scheduling, along with some kind of "public folder" support. (I put that in quotes, because "public folders" are gradually becoming a kind of generic term that means whatever the person using it wants it to mean. In my case, I'm thinking along Exchange lines.)

iCal Server is based on the CalDAV standard, a set of protocols designed to allow for Exchange/Domino-like calendaring functionality, without having to tie yourself to any specific product. (A full list of the various standards involved can be found at http://www.calconnect.org/calendaringstandards.shtml) To assist with that end, Apple has joined CalConnect, a consortium focusing on calendaring interoperability.

The elephant in the room of course is Microsoft, who is not a member of CalConnect. What is iCal Server's solution for Microsoft clients? Well, if you're talking about Outlook, then there's a number of people writing connectors for Outlook 2003 or so, and 2007 may not even need a special connector, or so I've heard. (Note, I'd love to get some real info on this.)

So for Outlook users, the, er, outlook is pretty good. But what about Entourage users? I know people don't believe me, but really, there are people who like using Entourage, and not just in an Exchange environment. They're Mac users, what about them? Well, the answer for that is murky. By murky, I mean nonexistent. There's nothing that I've found in either iCal Server, Entourage, or CalDAV that talks about Entourage beyond "well, they can read iCal files, right?."

While I understand that yes, Microsoft has not published their DAV implementation as well as they should, it seems kind of a silly move to not support Mac users running one of the only currently available groupware clients, especially when at least one company, Kerio, does support Entourage. I get that Apple may have issues with supporting Microsoft clients, but getting any company to not only sign off on a groupware infrastructure change, but also a desktop change en masse is not going to be easy, if its even possible in most cases. Aside from the client issues, iCal Server includes one feature that normally is the realm of high-end, and high-priced implementations, namely clustering. With iCal Server, you'll be able to have multiple front ends talking to the same shared storage on a SAN. Obviously, Apple is pushing Xsan, mostly because it's about the only SAN product with direct Mac support, (another sore spot with me, but that's another article), but I'd imagine that any SAN with direct Mac OS X Server support could be used here.

In fact, Apple is specifically mentioning using iCal Server in an Active Directory environment, where you'd use Active Directory for user management. Both of these are significant. The clustering shows that while Apple may not be an "Enterprise" company, they understand that reliability is important to their customers and that clustering, at least on this level should be available without needing to sell offspring. The Active Directory mention shows that Apple is getting that no, people are not in fact going to chuck out a functioning Active Directory infrastructure just because Mac OS X is cool.

If Apple can get a good answer for Microsoft clients on the Mac as well as Windows, then along with the lack of a CAL tax, and dirt-cheap clustering, iCal Server could be a huge win for them.

Wiki Server

While Calendaring is important, it is not the end-all and be-all of groupware and collaboration. Meetings only take you so far, and if everyone can't be in the room, you still need to be able to work together. To assist with that, Apple is implementing a workgroup-based Wiki Server in Leopard Server.

The idea of a Wiki is well known to most people thanks primarily to Wikipedia. The Leopard Server version will allow you to tie in Wiki functionality to Open Directory user and group management, along with Leopard's iCal server to create a top"notch collaboration environment. Apple is obviously paying close attention to usability, so that you can create Wiki sites and pages without needing to get into Wiki syntax. The Wiki Server will keep full histories of changes, so revision tracking will be simpler.

This will be a huge feature for anyone working on collaborative documents with a widespread team while avoiding application and document version issues. Tied in with the existing weblog implementation in Mac OS X Server, the Wiki Server will make basic collaboration work much simpler. Is it going to be the equivalent of Domino or Sharepoint? No, but it won't need to be, and it will definitely be cheaper.

64-Bit Capabilities

With Leopard and Leopard Server, the higher levels of the OS are going to be 64-bit. While this will not make a huge difference at the human level, it will be of great use to certain types of applications, most notably databases, such as MySQL which ships with Mac OS X Server. Apple has been shipping machines able to handle physical RAM sizes well outside of the 32-bit limit, but even as of Mac OS X 10.4, 64-bit support has been quite limited. With Leopard, the OS as a whole will be able to make more use of things like more than 4 GB of RAM.

While most applications don't care so much, databases get a direct benefit from more RAM. The more data they can keep in RAM instead of the hard drive, the faster they get. This will benefit not just MySQL, but Oracle and Sybase too. Other applications like Mathematica, IDL, MatLab, and others will be able to do real useful things with the increased functionality that Leopard's improved 64-bit support will bring.

File Services

Rumors of ZFS aside, Leopard is planning on implementing more security in NFS with direct support for kerberized NFSv3. NFS is without a doubt, long in the tooth, but it still works, and is used by thousands of companies, both in the .com and the .edu sectors. Increasing security, especially via Kerberos which makes single-signon much easier is always a good thing. Apple is also hinting at some improvements in the Mac OS X automounter, which would be joyously welcomed, especially by laptop users in directory systems, for whom automount has been a nigh-constant pain point. Web Services

It looks like Leopard Server may finally default to Apache 2 instead of 1 for the "standard" web server. That would be a good thing for quite a few people, particularly Subversion users. The addition of support for Ruby on Rails with Mongrel continues Ruby's march towards world domination, and gives the "what's the best language" wars another native player on Mac OS X Server.

QuickTime Services

Apple is shipping an updated, (finally!) QuickTime Streaming Server, (QTSS) in Leopard, along with the new Podcast Producer. With regards to QTSS, well, it's not been improved in any meaningful way in a dog's age, so almost anything is welcome. However, the Open Directory integration Apple talks about could be of use, especially if the Open Directory integration makes things like only allowing specific users or groups access to specific streams possible. That's something that quite a few people I know would love to have, and if it's in QTSS in Leopard, that would be a powerful incentive to upgrade. Podcast Producer helps automate the creation, uploading, and publishing of podcasts. Considering the number of businesses hopping on the podcast wagon, if Apple can really make this dead simple, it would be, along with an improved QTSS, yet another vector into more businesses.

iChat Server 2

The big one here in the business world is built-in server-level chat archiving. This is a huge one in the business world, and without it, you either had a hard time using iChat, or you had to do the (quite) painful dance of getting Bandersnatch to work with iChat. iChat Server 2 supports server federation, which will allow users on the iChat server to talk with other systems that support XMPP, such as Google Talk without having to log into Google Talk directly. Kerberos support is on tap, adding iChat to the ever-increasing single sign-on love in Leopard.

I don't mean to be blas about IM here. While much of the world looks at it as a toy, it is for me, and quite a few of my compatriots, a critical business tool. It offers faster, richer communication than email, it's easier to shoot files back and forth, it has, at least for me, none of the spam problems that are gradually breaking email, and it's dead simple to set up conference "calls" as needed. IM is rapidly becoming the best tool around for getting work done, and Apple improving the business usability of iChat Server is going to be an important sales tool for them.

Mail

Along with iCal Server, Mail in Leopard will support clustering via Xsan at no extra license fee. (Yes, you have to set up an Xsan system if you don't have one, so there's a bit of a "tax" there, but a good clustering setup begs for a SAN anyway.) When you talk about email, you're talking about a business, and even mission-critical service. I think it's safe to say that any company big enough to need its own email server has made email into a critical service. Even in a relatively small company, email going down for a few hours will cause real business problems. The classic way to deal with increased reliability is of course, clustering. Have multiple front ends talking to the same backend mail store. One front end goes down, you barely even notice. The problem with email clustering is that historically, it's either not cheap, (some implementations start at $50,000 US, and that's just for the license.), or easy to set up, or both. Apple enabling you to set up email clusters for $1,000 US or so per server? Dirt cheap. Really. Add into that the groupware improvements that Apple is putting into Leopard and you have a product that can do a lot of damage to the competition, in particular, Microsoft Small Business Server.

Oh, and there are improved vacation messages in the Leopard mail server, but those are evil, and I shall not speak of them further.

Xgrid 2

Xgrid is one of those odd technologies. For the people who need it and use it, it's invaluable. For everyone else, it's hard to even explain why it's cool. So if you don't need Xgrid, skip to the next section. For those of you who do, Apple is making some nice improvements to Xgrid, such as making it much easier to set up grids, and implementing a scoring system that helps assign jobs to the most appropriate nodes.

Open Directory 4

Open Directory4 includes support for cross-domain authentication. If this means that setting up Kerberos cross-realm authentication will get easier, I'm more than a little happy about that. The other big improvement here is the integration of RADIUS into Open Directory, a big win for anyone with large wireless networks. True, you could get OpenRADIUS to work with Open Directory in Mac OS X 10.4, but if Leopard gives you that integration without a ton of work, hey, I'm not going to complain.

What Does It All Add Up To?

Admittedly, this article spends a lot of time on the iCal Server, and barely any time on things like Open Directory 4. Not that the iCal server isn't important, but I'm limited by Apple's public information. On the iCal Server, they had a lot. On Open Directory4? Not so much.

I do think that it's obvious that Apple is making a lot of improvements in Leopard Server that will directly benefit the SMB market. No, Apple is not an "enterprise" company. They never really have been, and I doubt it's a market they want to get into ala Dell or Microsoft, or IBM. That's not to say that Leopard won't play well in the enterprise. On the contrary, I think it will have a lot of uses, even in big enterprisea. But I just don't see Apple wanting to make the changes that becoming an enterprise company would require.

That doesn't change the fact that Leopard Server is shaping up to be a great SMB product. For about a thousand bucks on existing hardware, or for the cost of an Xserve, you get a really solid server, able to support Web services, collaboration, groupware, IM, and file services. You can run it with its own directory service, or as part of an Active Directory implementation out of the box. It provides some features that due to pricing and/or setup requirements, have traditionally been reserved for "big" enterprises--in particular clustering of both email and calendaring servers.

Apple is taking items that either cost a lot of money, or need a lot of expertise, and selling them real cheap, and in a way that while you still need some basic networking knowledge and skills to set up, won't require the same levels as other systems. Will it let you gut a large-scale Active Directory/Exchange system? Of course not, but if you're a business with a hundred or so people, and you want to add services, increase reliability, and not go into deep debt to do so, Leopard Server is going to be a compelling option.