The Sametime Gateway
Based on a lengthy conversation at Ed's about the new Sametime Gateway release, Adam Gartenberg replied with a post of his own. Since he's the Offering Manager for Sametime, he is THE man to address some of the concerns from Ed's audience about the gateway. Since my reply is (predictably) too long to put post over there, it's time for some trackbacks.
1) Adam, you didn't really answer the first question you asked: "Why does the gateway require DB2?" This doesn't seem to be a hefty question to me. From what Chris Miller has described, the gateway actually surfaces your entire company mail directory as reachable addresses from services such as AIM and Yahoo. That's pretty cool, I must say, and would obviously require a data store to map user IDs to some arbitrary account created on the other service. I'm sure the gateway also allows for logging of all traffic that passes over it, which is similarly cool, and again obviously requires a data store. Lastly, there's probably some policy controls on who can use the gateway, and we have to have a place to track that, too.
What's wrong with Domino for these goals? Domino has a very powerful and capable directory system. It knows how to contains policy controls for services. And it has a *gasp* LOG, too! In fact, if a shop was using Domino as its email system, it would be fairly trivial to integrate IM logging directly into the mail facilities that already exist. The integrated Sametime client in Notes 7 ALREADY DOES THIS. Plus, at a server gateway level, you could integrate it with any mail journaling settings a customer might have, giving them a single, encrypted, text-searchable data store for all computer external communications! That sounds pretty useful to me.
So there must be something REALLY special that having a relational database does for the gateway product. I look forward to hearing your reply on what that is in your forthcoming complete answer.
3) (yeah, I'm doing this out of order) Adam, you mention my mention of Trillian as a comparison. I think you misunderstood my point here. Trillian isn't a gateway system. It's not a proxy -- I know that. My point in bringing up Trillian is that it's a product I can get for free with a negligible footprint that provides communication with a multitude of IM systems. It even allows me to maintain multiple, distinct accounts in simultaneous operations on the same IM platform (I can be logged into AIM under several accounts at once.) Does it integrate with my Notes client? Of course not. Does it provide for administration of policy controls for multiple users? In a sense, in that I can set different behaviors for different concurrent AIM logins. But my point in bringing it up is that it ALREADY did the bulk of what anyone needs an IM gateway for: translates between varying messaging protocols to allow for a single point of interface with multiple IM presences. Why are they able to do with such a simple architecture what seems to take IBM a battery of product platforms to accomplish?
To put it another way: Why do I need WAS to run a gateway product? It's an IM PROXY. Would MS tell me I need IIS to have Exchange send SMTP mail? (Maybe they would, but we'd all make fun of them for it.) What does a J2EE system have to do with IM translation?
2) Adam, how can you think it looks like anything other than IBM not addressing SMB customers in this release? In shops that aren't software companies, installing a web application server and a relational database system requires CONSIDERABLE investment. One of the great things about Domino is that it is SO easy to get running. And that's one of the reasons why MS has so much success in the smaller market. When you buy an MS Back Office server, you know that you're going to put in a CD/DVD and click on a series of buttons to get things running. IBM has LONG touted the simplicity of getting Domino upgraded vs. Exchange. Why does that suddenly not matter in the case of IM interoperability?
Look at it another way: How many WAS servers are there globally? How many DB2 servers are there globally? How many Domino servers are there globally? When you get the answer, pick your market.
(And for those who say "yeah, but there can be 20K users on a single DB2 server," my reply is "Yes, and that's at LARGE customers, not SMBs. Count server INSTANCES to see the difference in large vs. medium vs. small businesses")
Remember a while ago, during the "dark times," when IBM kept making public statements about killing Domino and forcing customers to move to WAS? It seemed that Notes/Domino was on a resurgence, and IBM finally understood the tremendous value of the platform. But maybe not. Maybe they just stopped TALKING about moving to WAS. And instead, with releases like this gateway, they're just going to force the shift WITH CODE.
That would be unfortunate. It would suck to see Damien proved right.
Comments
Posted by Steve Smillie At 09:58:33 AM On 12/08/2006 |
I wholly agree the IBM solution is a larger footprint, requires greater hardware and eats more local memory however, as I have said many times
Posted by Chris Miller At 09:29:22 PM On 12/09/2006 |
Posted by Ian Randall At 12:25:36 AM On 12/08/2006 |
{ Link }
Posted by Alan Bell At 05:29:17 AM On 12/09/2006 |
{ Link }
Posted by Adam Gartenberg At 03:41:36 PM On 12/12/2006 |
And I believe more firmly now than then that building Garnet into Domino would have created a frankenstein. But that wasn't my point in bringing it up.
Posted by Ed Brill At 10:32:38 AM On 12/08/2006 |
And, for the record, if this whole conversation is spin, let me add something... it's perfectly ok if the answer is "we already wrote all this stuff for some other purpose, and now we just plugged a Sametime element into what we'd already done."
Posted by Nathan T. Freeman At 10:00:11 AM On 12/11/2006 |
Posted by Ed Brill At 08:45:43 AM On 12/08/2006 |
Posted by Nathan T. Freeman At 10:46:38 PM On 12/07/2006 |
Posted by Adam Gartenberg At 11:01:00 PM On 12/10/2006 |
IBM is reading the LDAP source and taking the actual Internet addresses of employees out to the public system. As a long time user of Trillian Pro, I love the way it alows me to have mutiple presences in each protocol, not only for testing but for customers.
However, they are all unique plug-ins that require different logins for each system. Where Sametime Gateway does it on the server end for one name.
I do agree it needs to be an option of an appliance and simplified install however.
Posted by Chris Miller At 04:16:09 PM On 12/08/2006 |
Posted by Alan Bell At 05:19:43 AM On 12/09/2006 |
On the point that you could get the name nathan.freeman@lotus911 on AIM, that would only be true if someone else hadn't taken it.
There have been alot fo cases recently where MSN users that had take names for domains have been losing their name, even though they may have been using it quite happily for a few years. But that was because the person registering the name was a tad silly thinking that giving themselves a name that pointed to someone elses domain was somehow ok.
Carl.Tyler@lotus911.com
Posted by Carl Tyler At 08:30:38 PM On 12/13/2006 |
The difference is: they would do it in a small footprint that didn't require a J2EE server and a terabyte RDBMS.
I'll say it in a comment a little more bluntly than I would in the original post: they solved the hard problem in 8MB on a client system. Why can't IBM?
Yes, I know Trillian is a client, not a server. And I know that it doesn't include the administrative control that the Sametime Gateway does. What Trillian effectively points out is that IBM apparently needs a J2EE server and a terabyte RDBMS simply to auto-register users and track logs.
What else can I say? That simply doesn't speak well for IBM.
Posted by Nathan T. Freeman At 09:05:12 AM On 12/08/2006 |
Actually yes, that's been true since Exchange 2000.
As for your other comments, I realize that the Sametime Gateway and Activities are held out as the latest evidence of conspiracy. I can only tell you that there are more developers on Notes 8 than any other Notes release; that we have already approved a concept plan for the next Domino release after 8 (and interim work between major releases), and that more activity is underway (at least in part prompted by some of the blog conversations of the last few weeks).
I'd be careful with the assertion that there are more Domino servers globally than WAS or DB2. I don't know the sales figures for the other products, but I do know that DB2 is running almost every major bank or insurance company in the world, along with hefty amounts of government organizations.
I don't mean to ignore or diminsh your main point, which is that there's a tradition of administrative simplicity in Domino that needs to be extended to other Lotus products. Remember that we didn't get the four-step Domino install wizard until R4.6, Domino Administrator until R5, and are still getting more tools per release. I'm hoping that some of the blog world conversations will influence priorities for next versions of some of these products.
Posted by Ed Brill At 09:21:06 PM On 12/07/2006 |
I really don't get what the big deal about this "enterprise name" stuff is. While it's obviously a great thing, IBM didn't achieve it through great TECHNOLOGY. They achieved it through the BUSINESS RELATIONSHIP. After all, the only thing that keeps me from registering as nathan.freeman@lotus911 at AIM right now is that they'll only allow 16 characters in a name, and it can't include punctuation. But that's AIM's enforced limitation, not anything that has to do with how a proxy functions.
When I compare the gateway to Trillian, I'm not saying that Trillian could replace the gateway. I'm saying that the free product solved the hard problem in a tiny footprint. IBM's solution requires a world-class J2EE and a terabyte RDBMS. WHY IS THERE SUCH A BIG DIFFERENCE?
What is difficult to understand about my question, Chris? Auto-registration of users is not a massive technical accomplishment. Logging is not a massive technical accomplishment. Policy control is not a massive technical accomplishment. They're all pretty simple, and could easily be achieved in a Domino environment. It's dealing with the multiple protocols that's impressive. But Trillian does this with virtually no resources, while IBM's requiring a 4GB RAM server to do it.
WHY?
Posted by Nathan T. Freeman At 05:06:11 PM On 12/08/2006 |
Conspiracy, as in, "a combination of persons for a secret, unlawful, or evil purpose", is the wrong word and thought.
As an enormously large company (almost as big as HP!) IBM is a many legged animal (hmmm, somewhat like a caterpilar...) and has many things going on at once. Many of them are 100% dead-on in terms of direction. Some of them are not.
It seems that the 'dark ages' are behind us in a number of ways - IBM actually thinks the Notes client is cool!. In some ways they are not - You can use Domino as a non-portal web platform? You can build enterprise-class data driven applications on NSFs?
Reading between the lines here, it seems like in some important ways IBM gets that Notes & Domino are important, but not WHY. Damien K. does.
It's easy to talk about TCO, but that's where ND rocks. Yes, you can build Notes-like applications on a proprietry platform based on J2EE, but it's not cost effective when compared to what it takes to the same thing on with Domino. Domino - it's not just better, it's $marter and $weeter.
In my opinion, the future of Notes looks great. The future of Domino as it supports Notes is great, however the rest is open, but I think that these are some key things to consider:
Adding WAS (CE?) and integrating development on it seemlessly with Domino Designer in Eclipse,
Enhancing the whole HTTP side to allow easy CSS & AJAX driven design & development in DD,
Developing a data layer that allows use of Domino or DB2 for things like RTC & Activiy Explorer,
Reconciling within IBM that it's OK to develop cool WAS-based apps that store their data in NSFs.
Posted by Craig Wiseman At 10:41:22 PM On 12/07/2006 |
#3 - There are essentially two ways to provide interoperability - a multiheaded client (like Trillian) and a server-based approach (like we're doing). I'm not saying that the Trillian approach isn't appropriate for a set of users out there. What I am saying is that, setting aside our specific implementation for a minute, the businesses we work with (and especially the Public IM networks we are working with) by and large agree that a server-based approach is preferable for them over a multi-headed client. It's not that we couldn't have replicated what Trillian did using Sametime, it's that the administrators and IT departments we work with in our customers said they want the ability to centrally administer the connectivity. (And one clarification - I wasn't refering to a user administering their accounts, I was refering to the ability of administrators to control which users have the ability to connect to public IM networks.) Businesses like the idea of their users not needing to register personal IDs to be able to conduct business over Public IM networks. (If you're talking to your financial advisor about your retirement account, wouldn't you rather ping them at Bob.Smith@bank.com instead of BillyBob369@aol.com?) Again - I'm not saying it's right for everyone, but this is why we took the approach that we did.
#2 - Let's separate out for a minute intentions vs. delivery. I know what we delivered in the first release of the gateway is far from a one-click install. My main point was that we have not deselected SMB as a market; we do need to continue to do a better job of delivering solutions optimized for those customers.
Posted by Adam Gartenberg At 07:53:02 AM On 12/08/2006 |