Blogs

RedMonk

Skip to content

Open Source MQSeries, Science Experiments and Ruby on zSeries

James McGovern asks some questions on mainframes and open source.
 
Q. Does Ruby run on zOS?
A: Not that I know of. But you might look at NetRexx, which certainly does.
 
Q: Open source equivalents of RACF?
A: Not even sure what you mean by that, James. An equivalent?
 
Q: There are many folks on Wall Street that are working on an open source equivalent of MQ Series. Would be curious to know what industry analysts think of this effort? Would also love to know if it possible for a traditional end-user client of an analyst firm to create software and get it on a quadrant without having to become a software vendor?
A: I also talked to Coridan, which is another open source message queing effort, as per your earlier suggestion, and while I thought the MantaRay approach was potentially interesting, I am a little worried about who the target market is in both cases. Big Wall Street firms have casts of thousands that love to build IT stuff. They have people that like to go toe to toe with IBM and other major vendors, architecture for architecture, flexing technical pectorals. But the level of resource in financial services is hardly the enterprise norm. Banking is IBMland: it makes sense to put pressure on it. But MQSeries, now called WebSphere MQ, is pretty darned solid middleware. I do wonder sometimes about the balance between function and “non-proprietary” choices. Many SOA implementations are driving WebSphereMQ volumes, without requiring other WebSphere componentry. It is a very handy transport which covers an unbelievable number of platforms. It scales in a way JMS can’t. Without a vendor sugardaddy on board, what happens if banks cut the AMQ budget? What is the maintenance and governance model? I am all for enterprises contributing to open source, and it will be interesting to see if AMQ becomes, say, an Apache submission. Ron at Zapthink talks about open source message queing putting pressure on ESBs. I think of MQ as a transport not an ESB. In the end, I think I need to talk to AMQ and JP MorganChase to find out more. I have many questions before I can really say what I think of the effort.
 
AMQ might even act as leverage to push IBM towards thoughts of open sourcing the core WebSphereMQ messaging service. Its not as if customers wouldn’t still come to IBM for service and support and the rest of the stack… and who knows, IBM might even find one of its venerable middleware platforms winning new customers by lowering barriers to participation. IBM could then focus on selling the higher level software, for SOA Governance and so on. I wonder what the guys at Hursley think, not that they will likely say in public…
 
Disclaimer: IBM Software Group is a client.
Coridan isn’t. JP MorganChase isn’t.
James McGovern is a member of the RedMonk information community, althought is not currently a paying client. He is an architect at the Hartford. He has a challenging blog, as per the link above.
 

Categories: Uncategorized.

Comment Feed

14 Responses

  1. WebSphereMQ scales in a way JMS can’t?
    Do you still feel this? I’m with you on the number of supported platforms but, there are already some pretty big deployments of JMS in production and I have the feeling that this isn’t as true has it was some years ago. Do you have any data on deployments like the UK’s National Health System?

  2. On the mainframe there are only two known security products. One is RACF and the other is TOP Secret. Would have figured that someone would have been noodling an open source competitor in this space and put in features to make them more interoperable with the rest of the world.

    For example, imagine if an equivalent could consume SPML messages from an identity management platform or accept SAML…

  3. One other thought, I have been on a rant against the Ruby community but have been noodling working with them to port Ruby to Z/OS if I could find others that would be willing to participate.

    If we go through this effort, would other folks actually use it?

  4. Three security products, actually: RACF, TopSecret, and ACF2. The first is from IBM, and the latter two are from CA.

  5. Jaime, I’m with James on this (re: MQ scalability).

    That said, it’s confusing to me when it’s phrased as MQ v. JMS. MQ can be (and often is) a JMS provider and is then written as JMS/MQ or MQ/JMS (or maybe with hyphens). So MQ versus JMS is confusing because JMS is an API (programming interface), and the MQ transport does provide the JMS API.

  6. Do you see any reason why the Ruby community wouldn’t want to support a port to Z/OS?

    AnonymousMarch 23, 2006 @ 1:03 pmReply
  7. There are three generally available security systems for z/OS. RACF from IBM plus ACF/2 and TopSecret from CA.

    Security works on z/OS by being embedded (and unavoidable) at control points in the OS itself, as well as having software-visible interfaces usable by properly configured applications such as database managers.

    Unless the open source community can get their security effort embedded to a similar degree in Linux and accepted as a base part of the Linux distribution, it is hard to see how such an effort could succeed technically or provide value.

    Chris CraddockMarch 24, 2006 @ 11:46 amReply
  8. Thimothy: OK, So, do you have any more or less recent Data on scalability of both (errr) “Products”?
    I’m not asking just to rant or something, I really would love some more Data on this issue (either way it goes).

  9. BTW folks often use the AMQ acronym to refer to ActiveMQ :). The JPMorganChase initiative for a wire level protocol for MOM is called AMQP (the P on the end stands for Protocol). I worked on that over a year ago – I’ve no idea if its ever going to be public or not.

    Also there is already an Apache submission now for an open source alternative to MQSeries together with 2 open wire protocols…

    http://incubator.apache.org/activemq/

    The 2 open wire protocols that ActiveMQ supports are either a fast binary (& harder to implement) OpenWire format…

    http://incubator.apache.org/activemq/OpenWire

    or the really simple & easy to implement text based Stomp protocol (think HTTP-like but working for MOM)…

    http://incubator.apache.org/activemq/Stomp

    so now there are clients in C, C++, C#, Ruby, Python, Perl, PHP, Pike etc. Plus any other messaging system can easily support these 2 protocols if they wish.

    Disclaimer: I’m a developer on ActiveMQ and I also worked with JPMorganChase to help define the AMQP protocol.

  10. I do agree with your reading of the open source messaging market, and I believe that this market will mirror the trends in the application server market. Here we are seeing a bifurcation of the market into the high-end (mission critical, high availability/reliability/scalability) and the low-end (“good enough”). The high-end requirements are being met by WebSphere Application Server ND/XD, and the low-end is being served by our open source-based application server, WebSphere Application Server Community Edition. Similarly, Websphere MQ serves the high-end market, while the low-end market is being tackled by a number of open source implementations. Large financial firms that bet their business on industrial strength messaging are better served by WebSphere MQ.

    As for IBM’s entry into open source in this area, we see this as an opportunity to reach new customers. For instance, we already use ActiveMQ today to implement JMS in WebSphere Application Server Community Edition. ActiveMQ is one of the leading open source implementations of JMS, and also implements the AMQ protocol that you mentioned in your blog. ActiveMQ is in the process of being accepted into the Apache Software Foundation (eventually to become a subproject of Apache Geronimo) – so to your point of whether there will be an Apache submission here, that is already in progress.

    Alice Chou
    WebSphere Marketing | IBM Software Group

    Alice ChouMay 12, 2006 @ 10:58 amReply
  11. This article has more detail about the new AMQP protocol. The AMQP specs appear to be available online. No mention if there is software or just specs:

    http://news.taborcommunications.com/msgget.jsp?mid=698847

  12. And then there were four: http://www.jmesoftware.com/deadbolt/.

    NB Barry Schrager is one of the initials in SKK, the creators of ACF2 (Klemens and Krueger being the others).



Some HTML is OK

or, reply to this post via trackback.

Continuing the Discussion

  1. McGovern Comes out Swingin’ on Ruby in the Enterprise

    It’s early Sunday morning, so my brain-processing isn’t quite hiked up enough to write anything cogent about James McGovern’s post yesterday on Ruby: “Ruby, to me feels like a trainwreck waiting to happen. So lets list out reasons why Ruby…