CICS Blog
August 15, 2005
This is funny!
My wife is into language, grammar and spelling (used to be a Technical Writer and Editor at IBM), so I get picked up regularly on my speling mistooks.
Being known for this, she was sent a link to the Dialectizer, and I found pointing it at this blog very amusing!
Although I don't come from strictly the right part of London to be a Cockney, I think that you might agree that the Dialectizer does a better job of rendering my true "style" than I can myself! If you're coming to SHARE, then you'll be able to let me know if you like this posh version or the dialect version.
(The other dialects on the site are funny too, but Cockney justs seems too funny!)
Posted by Ian Mitchell at 08:40 AM
| Comments (0)
August 10, 2005
On the art (and beauty) of software implementations
I stumbled on The fine art of computer programming and it links to Mob Software: The Erotic Life of Code. I found both of these articles fascinating reading - style does count.
Posted by Ian Mitchell at 02:41 PM
| Comments (0)
August 09, 2005
CICSPlex/SM and its Web User Interface
As many of you will already know, it's been announced that the TSO End-user-interface (TSO EUI) will be removed in the next release of CICS Transaction Server.
We know that many of the current CICPlex/SM users are concerned about switching over to the Web User Interface (WUI), and I'd welcome any specific comments one this, either here or via email.
Dibbe Edwards, the CICS Products Development Director, explained at the T&M in Orlando that we want to work with customers to make sure we're all sufficiently happy before taking TSO EUI away.
I've heard of a number of people getting concerned that a WUI infrastructure is in some way more vulnerable to disruption than the TSO components we know and love. I'd like to explain why I don't believe that ought to be a concern that enters the debate.
When we say it's a "Web user interface", we're really saying that it's an interface for users via a web browser. The web browser replaces the TSO terminal and provides a more graphical, familiar experience for people new to the platform, and more effective facilities for the advanced user. But there is no other significant change in the architecture and components used to support this. There is no 'real' webserver involved. The browser connects directly to a CICS region that serves up the HTML using the CICS Web support.
So, broadly speaking, the browser replaces the 3270 terminal (most likely an emulator - who's still running 'real' 3270s anymore?), HTML replaces 3270 (which probably goes over TCP/IP via TN3270 anyway), the CICS WUI Server replaces the TSO CAS, and the requests are still carried out in a combination of the CMAS and agent code in the MASes.
This means that you should not have any worries that any of this infrastructure is vulnerable to the sorts of common issues suffered by other "web apps" - there's no Apache, WAS, or other 'commodity' server involved.
July 21, 2005
CICS TS v3.1 and Enterprise Service Buses
"CICS TS and the ESB" is a topic I'm asked about a lot.
Firstly, what's the ESB? You can read much more about the official IBM definition and which IBM products embody the concept here.
That description leads with "The enterprise service bus (ESB) is an architectural pattern that offers a comprehensive, flexible and consistent approach to integration." This is all fine and dandy, and is a pre-req to answering the question about CICS TS and the ESB.
I work quite closely with some of the leading architects of IBM's vision of the ESB, which has helped me to gain a perspective on CICS TS and the ESB. I have found that it's all too easy to get too focused on the bus at the expense of the larger picture. Remember that the bus is an "approach to integration" - integrating what? In my ESB terminology the things being integrated are "endpoints". And CICS TS v3.1 aims to be a first class endpoint on an ESB.
IBM's ESB strategy is largely founded on using Web services as a common set of architectural standards. Notice I said "Web services" and not "SOAP". It is more important to achieve commonality of sevice descriptions via WSDL, for example, than to necessarily assume a service will be invoked via SOAP/HTTP. WSDL (and WS-Policy) provide an opportunity for different mechanisms to be used to connect a service requester and provider. CICS TS has aimed for good support of SOAP/HTTP and SOAP/MQ (as defacto standard bindings) with v3.1, but there are additional opportunities we may consider for the future.
It is the Bus's role to step in and glue the service provider and requester together when a more direct connection is not possible, or additional function is required. But for a CICS TS ESB endpoint, SOAP/HTTP and SOAP/MQ are two good options for "on-ramps" to an ESB. Alternative deployments of CICS TS with an ESB include using the JCA facilities provided by the CICS Transaction Gateway, although this only supports a scenario with CICS TS as a service provider. We are seeing many cases where CICS TS customers are demanding CICS TS applications to act in the service requester role too, and v3.1 supports that via the new EXEC CICS INVOKE WEBSERVICE command.
The different perspective that I have on the ESB is gained by being the architect of an endpoint rather than the bus itself. What we need to ensure is that CICS TS keeps up with the evolution and wider adoption of ESBs, but always keep in mind that its role is that of endpoint. Perhaps my slogan ought to be "the endpoint is the whole point" - hey, perhaps that should apply to the ESB too!
Posted by Ian Mitchell at 11:42 AM
| Comments (2)
July 06, 2005
Is Web services in CICS TS v3.1 simple?
Following on from that last entry (I didn't want to cover two topics in one entry, it'd only lead to confusion in any comments you make!) where I mentioned simplicity of Web services in CICS TS v3.1.
I had an interesting experience at a customer a few weeks ago - I was enthusiastically explaining all the ways in which we've added features to make the processing of Web service requests highly configurable (URIMaps, Pipeline definitions, Web service RDO, WSBind files, etc), but at same time simple (ie deployable "out-of-the-box" with sample definitions). At the session we'd encouraged application developers and architects, as well as the sysprogs, to attend. After a while, a member of the audience raised his hand to ask a question - "I'm an application developer, writing COBOL business logic - what do I have to do differently to enable Web services?" Well, what a good question that was! It's all too easy to focus on the new mechanisms that enable Web services, and forget that the point is to enable use of business logic. So, to answer the guy's question, I said "if you're writing pure business logic, of an appropriate granularity, with a commarea interface, then you can continue doing just that."
Now, this customer has some relatively mature architectural standards for their CICS applications, which meant that they've been "SOA-style" for a number of years. So, in this case, they have a very good chance of exploiting Web services with little or no change in what and how the application developers go about their business.
Posted by Ian Mitchell at 09:07 AM
| Comments (5)
Last Week's T&M Conference in Orlando
I was not at the T&M last week in Orlando, but from what I've heard from some of the Hursley contingent, it seems that the sessions on 3.1 features were very popular - especially Web services. I get the impression that some of you did not quite believe that enablement could be quite as simple as we've tried to make it, but I heard that Mark Cocker's workshop did a very convincing job.
Anyone else care to comment on the conference overall?
I expect to be at SHARE in Boston , so if you're there too, we can chat about this blog face to face.
Posted by Ian Mitchell at 08:19 AM
| Comments (1)
June 30, 2005
This blog vs CICS-L
Roland Schiradin has asked a very reasonable question in a comment to my opening entry, namely how does this blog relate to CICS-L?
This blog has a very different purpose to CICS-L, indeed blogs are very different to listserv "thingies" - both in philosophy and implementation.
In contrast to CICS-L, the direction this blog takes is more under the control of Arthur, Keiron and myself - it's our blog, but we hope what we say in it will be interesting and relevant to you, and we want feedback. The subjects that come up on CICS-L are driven by the subscribers and posters. CICS-L serves a very valuable need for self-help and community communications ("how do I...?", "has anyone done...", etc) - the sort of topics I anticipate here in the blog will be more of the flavour of "did you know that....", "I saw this the other day and wondered how it relates to CICS...", and, within the bounds of what won't get me fired, "what would you think about CICS doing....".
Plus, and I hope you'll forgive a little rant, from the practical point of view, I can't deal with all the formatting and "out of office" rubbish that you have to deal with on CICS-L. I've got better things to do that filter out all the HTML junk that appears in the postings.
Posted by Ian Mitchell at 08:34 AM
| Comments (0)
Blogging - tips and techniques
Well, I must say there's a bit more to this blogging lark than I'd imagined.
Sorry to those of you who suffered the 'registration required' error - I've now pointed to WebSphere.org as the place to start to register.
The error about "comment templates" when you made a comment even when you are registered - has been fixed!
The comments are getting through and I get to "approve" them before they appear on the site - wow, such power! But unless you're abusive then I'll not be censoring anything.
Anyway - it'd like to invite those of you reading this who are more experienced in blogging, to consider dropping a line in a comment with any tips and techniques you think might be helpful.
Posted by Ian Mitchell at 08:21 AM
| Comments (0)
June 17, 2005
Ian Mitchell - who's that?
I'd had an idea of starting an 'unofficial' CICS blog a couple of weeks before being asked to participate here - but now it comes down to it, starting off a blog seems quite intimidating!
But there, I've done it!
So since this is not my 'unofficial' CICS blog that I'd thought of doing - what is it? Well, it's 'official' in terms of the activity being blessed and encouraged by 'the management' (IBM that is, not SWMBO!), but the content should be regarded as distinctly 'unofficial'! As it's a blog, it's me - sharing some thoughts on what I'm up to in my role at IBM.
Anyway, to answer the question that I posed at the top...
Currently, I'm the Architect for CICS Transaction Server, and it's CICS that I intend to focus on in this blog. But the topics will range around many areas related to CICS.
Anyone who's met me at any of the conferences will already realise that I should not be confused with the other Ian Mitchell you'll find in Google!
I work at the Hursley Lab in the UK - the village is described in Wikipedia, and, of course, there is an official IBM page for the Lab.
I've actually worked at Hursley on CICS 'forever' - that is the only gainful employment has been here, working on CICS since 1987. Some might think "you sad article", but it's been, generally, an immensely interesting, satisfying and fun. I doubt we'll get very far in this blog without some mention of some of the fine friends that I've worked with over the years.
Being the Architect is somewhat daunting, but if we get some more releases like 3.1 out the door, then I'll be more than happy. What the entire CICS team has achieved with the latest release is truly outstanding. Of course, the team at Hursley is central to that, but there are many groups and individuals throughout IBM that make significant contributions to each release. And this time around, I would also include many of the customers as part of the team that has given us 3.1. I, and many other people from Hursley had the opportunity to discuss our ideas for what to put in 3.1, and how to do it, with many customers. I intend to discuss many of the features of 3.1 in blog entries to come - and I'd love it if you discussed them here too - go on, push that button and write a comment!
(You'll need to register - go to the WebSphere.Org page and follow the login link in the menubar at the top.)
Posted by Ian Mitchell at 12:53 PM
| Comments (7)
May 27, 2005
CICS, SOA, WSAA and Relativity..
Hello, you can tell I'm English already can't you?
Just for a change I've been talking to retailer customers about the new CICS, (If I have to say CICS TS 3.1 you're probably reading the wrong BLOG) I've been doing this for 10 years (First for CICS distributed, then MQ/WBI and now back to big CICS) so I should be used to it, but taking the "new baby" out to the people who pay for it is always unnerving. But, so far, so good, WebServices is "in" and now people look at my SOA foil and the stuff on XML and go "Ah, WebServices!"...
There’s a load of new function in CICS and one or two customers are starting to see how Assett Analyser works with Web Services especially since the Relativity deal got announced. Yet another case of stealth marketing I think. “AA” with 3.1 could be finally answer the question "Where do I get time to make the services?"
Account teams (that’s IBMers) tell me they don’t like what we did to COBOL: we finally took OS/VS support out in 3.1 ("they" told us it was long dead). Now account teams tell me there still seems to be a lot of OS/VS stuff out there and if the source isn’t lost it's certainly proving elusive. And COBOL programmers are a bit thin on the ground, I pass this stuff back to Marketing and development, but if you have stories of how it’s proving a problem let me know..
Now I'm off to Milan and then Prague to talk to customers, I'll let you know what was the "hot topic" next time..
Posted by Keiron Casey at 09:37 AM
| Comments (4)
This is funny!
My wife is into language, grammar and spelling (used to be a Technical Writer and Editor at IBM), so I get picked up regularly on my speling mistooks.
Being known for this, she was sent a link to the Dialectizer, and I found pointing it at this blog very amusing!
Although I don't come from strictly the right part of London to be a Cockney, I think that you might agree that the Dialectizer does a better job of rendering my true "style" than I can myself! If you're coming to SHARE, then you'll be able to let me know if you like this posh version or the dialect version.
(The other dialects on the site are funny too, but Cockney justs seems too funny!)
On the art (and beauty) of software implementations
I stumbled on The fine art of computer programming and it links to Mob Software: The Erotic Life of Code. I found both of these articles fascinating reading - style does count.
CICSPlex/SM and its Web User Interface
As many of you will already know, it's been announced that the TSO End-user-interface (TSO EUI) will be removed in the next release of CICS Transaction Server.
We know that many of the current CICPlex/SM users are concerned about switching over to the Web User Interface (WUI), and I'd welcome any specific comments one this, either here or via email.
Dibbe Edwards, the CICS Products Development Director, explained at the T&M in Orlando that we want to work with customers to make sure we're all sufficiently happy before taking TSO EUI away.
I've heard of a number of people getting concerned that a WUI infrastructure is in some way more vulnerable to disruption than the TSO components we know and love. I'd like to explain why I don't believe that ought to be a concern that enters the debate.
When we say it's a "Web user interface", we're really saying that it's an interface for users via a web browser. The web browser replaces the TSO terminal and provides a more graphical, familiar experience for people new to the platform, and more effective facilities for the advanced user. But there is no other significant change in the architecture and components used to support this. There is no 'real' webserver involved. The browser connects directly to a CICS region that serves up the HTML using the CICS Web support.
So, broadly speaking, the browser replaces the 3270 terminal (most likely an emulator - who's still running 'real' 3270s anymore?), HTML replaces 3270 (which probably goes over TCP/IP via TN3270 anyway), the CICS WUI Server replaces the TSO CAS, and the requests are still carried out in a combination of the CMAS and agent code in the MASes.
This means that you should not have any worries that any of this infrastructure is vulnerable to the sorts of common issues suffered by other "web apps" - there's no Apache, WAS, or other 'commodity' server involved.
CICS TS v3.1 and Enterprise Service Buses
"CICS TS and the ESB" is a topic I'm asked about a lot.
Firstly, what's the ESB? You can read much more about the official IBM definition and which IBM products embody the concept here.
That description leads with "The enterprise service bus (ESB) is an architectural pattern that offers a comprehensive, flexible and consistent approach to integration." This is all fine and dandy, and is a pre-req to answering the question about CICS TS and the ESB.
I work quite closely with some of the leading architects of IBM's vision of the ESB, which has helped me to gain a perspective on CICS TS and the ESB. I have found that it's all too easy to get too focused on the bus at the expense of the larger picture. Remember that the bus is an "approach to integration" - integrating what? In my ESB terminology the things being integrated are "endpoints". And CICS TS v3.1 aims to be a first class endpoint on an ESB.
IBM's ESB strategy is largely founded on using Web services as a common set of architectural standards. Notice I said "Web services" and not "SOAP". It is more important to achieve commonality of sevice descriptions via WSDL, for example, than to necessarily assume a service will be invoked via SOAP/HTTP. WSDL (and WS-Policy) provide an opportunity for different mechanisms to be used to connect a service requester and provider. CICS TS has aimed for good support of SOAP/HTTP and SOAP/MQ (as defacto standard bindings) with v3.1, but there are additional opportunities we may consider for the future.
It is the Bus's role to step in and glue the service provider and requester together when a more direct connection is not possible, or additional function is required. But for a CICS TS ESB endpoint, SOAP/HTTP and SOAP/MQ are two good options for "on-ramps" to an ESB. Alternative deployments of CICS TS with an ESB include using the JCA facilities provided by the CICS Transaction Gateway, although this only supports a scenario with CICS TS as a service provider. We are seeing many cases where CICS TS customers are demanding CICS TS applications to act in the service requester role too, and v3.1 supports that via the new EXEC CICS INVOKE WEBSERVICE command.
The different perspective that I have on the ESB is gained by being the architect of an endpoint rather than the bus itself. What we need to ensure is that CICS TS keeps up with the evolution and wider adoption of ESBs, but always keep in mind that its role is that of endpoint. Perhaps my slogan ought to be "the endpoint is the whole point" - hey, perhaps that should apply to the ESB too!
Is Web services in CICS TS v3.1 simple?
Following on from that last entry (I didn't want to cover two topics in one entry, it'd only lead to confusion in any comments you make!) where I mentioned simplicity of Web services in CICS TS v3.1.
I had an interesting experience at a customer a few weeks ago - I was enthusiastically explaining all the ways in which we've added features to make the processing of Web service requests highly configurable (URIMaps, Pipeline definitions, Web service RDO, WSBind files, etc), but at same time simple (ie deployable "out-of-the-box" with sample definitions). At the session we'd encouraged application developers and architects, as well as the sysprogs, to attend. After a while, a member of the audience raised his hand to ask a question - "I'm an application developer, writing COBOL business logic - what do I have to do differently to enable Web services?" Well, what a good question that was! It's all too easy to focus on the new mechanisms that enable Web services, and forget that the point is to enable use of business logic. So, to answer the guy's question, I said "if you're writing pure business logic, of an appropriate granularity, with a commarea interface, then you can continue doing just that."
Now, this customer has some relatively mature architectural standards for their CICS applications, which meant that they've been "SOA-style" for a number of years. So, in this case, they have a very good chance of exploiting Web services with little or no change in what and how the application developers go about their business.
Last Week's T&M Conference in Orlando
I was not at the T&M last week in Orlando, but from what I've heard from some of the Hursley contingent, it seems that the sessions on 3.1 features were very popular - especially Web services. I get the impression that some of you did not quite believe that enablement could be quite as simple as we've tried to make it, but I heard that Mark Cocker's workshop did a very convincing job.
Anyone else care to comment on the conference overall?
I expect to be at SHARE in Boston , so if you're there too, we can chat about this blog face to face.
This blog vs CICS-L
Roland Schiradin has asked a very reasonable question in a comment to my opening entry, namely how does this blog relate to CICS-L?
This blog has a very different purpose to CICS-L, indeed blogs are very different to listserv "thingies" - both in philosophy and implementation.
In contrast to CICS-L, the direction this blog takes is more under the control of Arthur, Keiron and myself - it's our blog, but we hope what we say in it will be interesting and relevant to you, and we want feedback. The subjects that come up on CICS-L are driven by the subscribers and posters. CICS-L serves a very valuable need for self-help and community communications ("how do I...?", "has anyone done...", etc) - the sort of topics I anticipate here in the blog will be more of the flavour of "did you know that....", "I saw this the other day and wondered how it relates to CICS...", and, within the bounds of what won't get me fired, "what would you think about CICS doing....".
Plus, and I hope you'll forgive a little rant, from the practical point of view, I can't deal with all the formatting and "out of office" rubbish that you have to deal with on CICS-L. I've got better things to do that filter out all the HTML junk that appears in the postings.
Blogging - tips and techniques
Well, I must say there's a bit more to this blogging lark than I'd imagined.
Sorry to those of you who suffered the 'registration required' error - I've now pointed to WebSphere.org as the place to start to register.
The error about "comment templates" when you made a comment even when you are registered - has been fixed!
The comments are getting through and I get to "approve" them before they appear on the site - wow, such power! But unless you're abusive then I'll not be censoring anything.
Anyway - it'd like to invite those of you reading this who are more experienced in blogging, to consider dropping a line in a comment with any tips and techniques you think might be helpful.
Ian Mitchell - who's that?
I'd had an idea of starting an 'unofficial' CICS blog a couple of weeks before being asked to participate here - but now it comes down to it, starting off a blog seems quite intimidating!
But there, I've done it!
So since this is not my 'unofficial' CICS blog that I'd thought of doing - what is it? Well, it's 'official' in terms of the activity being blessed and encouraged by 'the management' (IBM that is, not SWMBO!), but the content should be regarded as distinctly 'unofficial'! As it's a blog, it's me - sharing some thoughts on what I'm up to in my role at IBM.
Anyway, to answer the question that I posed at the top...
Currently, I'm the Architect for CICS Transaction Server, and it's CICS that I intend to focus on in this blog. But the topics will range around many areas related to CICS.
Anyone who's met me at any of the conferences will already realise that I should not be confused with the other Ian Mitchell you'll find in Google!
I work at the Hursley Lab in the UK - the village is described in Wikipedia, and, of course, there is an official IBM page for the Lab.
I've actually worked at Hursley on CICS 'forever' - that is the only gainful employment has been here, working on CICS since 1987. Some might think "you sad article", but it's been, generally, an immensely interesting, satisfying and fun. I doubt we'll get very far in this blog without some mention of some of the fine friends that I've worked with over the years.
Being the Architect is somewhat daunting, but if we get some more releases like 3.1 out the door, then I'll be more than happy. What the entire CICS team has achieved with the latest release is truly outstanding. Of course, the team at Hursley is central to that, but there are many groups and individuals throughout IBM that make significant contributions to each release. And this time around, I would also include many of the customers as part of the team that has given us 3.1. I, and many other people from Hursley had the opportunity to discuss our ideas for what to put in 3.1, and how to do it, with many customers. I intend to discuss many of the features of 3.1 in blog entries to come - and I'd love it if you discussed them here too - go on, push that button and write a comment!
(You'll need to register - go to the WebSphere.Org page and follow the login link in the menubar at the top.)
CICS, SOA, WSAA and Relativity..
Hello, you can tell I'm English already can't you?
Just for a change I've been talking to retailer customers about the new CICS, (If I have to say CICS TS 3.1 you're probably reading the wrong BLOG) I've been doing this for 10 years (First for CICS distributed, then MQ/WBI and now back to big CICS) so I should be used to it, but taking the "new baby" out to the people who pay for it is always unnerving. But, so far, so good, WebServices is "in" and now people look at my SOA foil and the stuff on XML and go "Ah, WebServices!"...
There’s a load of new function in CICS and one or two customers are starting to see how Assett Analyser works with Web Services especially since the Relativity deal got announced. Yet another case of stealth marketing I think. “AA” with 3.1 could be finally answer the question "Where do I get time to make the services?"
Account teams (that’s IBMers) tell me they don’t like what we did to COBOL: we finally took OS/VS support out in 3.1 ("they" told us it was long dead). Now account teams tell me there still seems to be a lot of OS/VS stuff out there and if the source isn’t lost it's certainly proving elusive. And COBOL programmers are a bit thin on the ground, I pass this stuff back to Marketing and development, but if you have stories of how it’s proving a problem let me know..
Now I'm off to Milan and then Prague to talk to customers, I'll let you know what was the "hot topic" next time..
