Blog
You and i

Steve Will

Steve Will
Bio


Bookmark and Share

April 2014

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30      

04/08/2014

Announcing IBM i 7.1 TR8

Ibm i smallerToday, April 8, is announcement day for IBM i 7.1 Technology Refresh (TR) 8. That’s right, we are now the proud parents of eight semi-annual collections of new functions which have been added to the most powerful release we’ve ever shipped. When we shipped 7.1 four years ago, launching our new strategy of focusing on mid-release delivery of new function, we certainly expected it would provide a useful, non-disruptive means for our clients to adopt new technology. It has done that. What’s been particularly satisfying is to see the rate at which TRs are adopted.

 When the first TRs came out, of course, fewer people had moved to the IBM 7.1 release. So we knew they would be adopted by fewer people than later TRs. This was expected. What we didn’t know how to predict, though, was how fast people would adopt them. In those first couple of TRs, the adoption rate was steady, but low–well under 1,000 per month. For the more recent TRs, within the first couple of days after GA, we have thousands who have downloaded the TRs. And the rate from there is very steep.

The high rate of download demonstrates a few things:

  • There are a great many clients on 7.1, and that number is still growing steadily. We know this from other sources, of course, but TR adoption data is actually the fastest and easiest way for me to see the rate. We still have to extrapolate a bit—not every 7.1 customer adopts TRs, and among those who do, they don’t all adopt every one of them.
  • The community, as a whole, trusts the quality of the TRs. If they didn’t, I would expect to see a slow ramp-up to adoption of each new TR. Instead, the rate of adoption is increasing.
  • The word about TRs has reached a “critical mass” of customers and business partners. When we first introduce TRs, we realized we were going to need to spread the word about them. Our customer set had been conditioned for decades to think of PTFs as almost exclusively a “defect fix” mechanism. For this reason, many of them had rules about PTF application which slowed down adoption of TRs, which are delivered as a PTF Group together with a Cumulative PTF package. But now it is clear that most clients recognize the difference between PTFs which are “just fixes” and the TR deliveries.

So, what’s in the latest TR8 announcement? Several things, actually. There are new native attachment options (“native” meaning they do not require VIOS) related to 16 GB Fibre Channel and some SAS drawers. (That’s as much “hardware” as you’re ever likely to get from this blog author!) The WebSphere Liberty profile is now the base for the IBM i Integrated Application Server. And of course, DB2 continues to satisfy requests from users and ISVs, particularly in the areas of performance analysis and database engineering. One quick note, though – the actual GA for TR8 content is June 6, and the GA date for the other functions announced with the TR (such as the DB2 functions) is listed in the information for those functions. Check out the details on developerWorks, and in the several blogs and articles our IBM i team writes or helps produce.

The success of our Technology Refresh strategy has been quite satisfying, and the strategy will continue. But, as I have said before, we definitely still need major releases. So watch this space as April continues. There just might be more news before the month is over.

03/13/2014

IBM i 7.1 Knowledge Center, Redbooks and developerWorks

Three big items in the world of IBM i information are now available for public consumption, so it’s worth telling you about them. I’ll give you quick notes and links for each.

The IBM Knowledge Center I talked about last month is now live.

The link is http://www-01.ibm.com/support/knowledgecenter/ssw_ibm_i/welcome

This site is the replacement for the Information Center. It has all the information about our supported releases, and when 7.2 gets announced, you will find the technical reference material there.

Knowledge Center Live

Redbook CoverThe updated IBM i 7.1 Redbooks publication has been released in draft form. This document has been updated to contain technical information about content through IBM i 7.1 Technology Refresh 7. This will be the final update for this Redbooks publication, because we’ll start planning for a 7.2 Redbooks publication when the first of the TRs for that release becomes available. 

The link is http://www.redbooks.ibm.com/Redbooks.nsf/RedpieceAbstracts/sg247858.html?Open

Last, but not least, the IBM i zone in developerWorks has a new look. The entire developerWorks library is being relaunched using the IBM Connections product.

The link is https://www.ibm.com/developerworks/ibmi/

DeveloperWorks Connections

You’ll note that the link to the Technology Updates link, which takes you to TR information, is now a nice graphic of an arrow, on the right side of the page.

Things are getting exciting this late winter as we move toward the IBM i 7.2 announce, and it’s great to get these three things out ahead of that event, so they don’t get lost among all the other hype. The teams responsible for these hope you find them useful.

 

 

02/24/2014

DB 102: Database Orientation - Row vs. Column

Are you a database (DB) professional? If so, you can probably skip the blog this time.

Are you an IT professional who doesn’t get into the details of DBs? Then this blog is probably for you.

Are you someone other than an IT professional? Then I think you might have the wrong blog. Or you’re a member of my family. In either of those cases, thanks for the thought, but you should probably go read something else.

OK, I think I’ve reduced the audience now to people who might care about today’s topic. See, if you are a non-DB IT professional, you know that DBs are important, and you probably wish you knew a little more about one of the hot DB topics being discussed in IT these days.

That’s kind of where I was a while back. I freely admit to most anyone that, while I have worked on many (many!) parts of the IBM i operating system, I have never been a developer in the DB2 area. So I regularly need to ask some of the smart people on the DB2 team to give me some semi-advanced lessons on databases. In particular, I recently needed to hear about why so much of the industry is talking about “columnar” databases. So, I went to chat with Mark Anderson. Mark, as many of you will know, is the Distinguished Engineer who is the Chief Architect of DB2 on i. Today, I thought I’d share some of what I learned.

Many database conversations center on performance, and that’s true with our topic today. People are talking about the performance of the newer column-oriented DBs, in comparison to the traditional row-oriented DBs.

As their names imply, there is a difference between how the data is stored in each of these DB architectures. In row-oriented DBs, each piece of data in a row is put on disk very close to each other piece in that row. Typically, you can think of it as if a row is a continuous set of bytes all sitting on disk. What’s stored in a row? Well, if you remember from DB 101 (you did all take the intro DB course, right?) a row typically has all the information about some entity. For purposes of this discussion, let’s say a row contains information about a customer. Then all of the attributes of a single customer are stored together on disk. Each of those attributes is a column in the virtual table (or array) we’re storing. I think this will go better with an example.

DB Example - Customer Table

On disk, you have a large number of pages storing all this data. Let’s say that each row in the DB is large enough to take up one page of storage. In this case, that’s one page for each customer. OK, so given the example above, in a row-based DB, all of Alice’s information is stored in one page. If your application deals with getting information about Alice, or updating information about Alice, then all you need to have in memory at one time is that single page.

DB Example - Row-Based

Though I haven’t said it yet, you can probably imagine that in a column-based DB, all the data in a column is stored together. Again, let’s say in our example, a column takes a page. So, for example, all of the ZIP codes would be on one page. And all the values in “2013 Total Order” would be on another page.

DB Example - Column-Based

(And hey, all you DB experts are supposed to have stopped reading, so I don’t want any smart comments about the design of the specific customer table above. I’m not a DB architect! This is a teaching example.)

OK, so now we can get to the point.

All databases (all computing, in fact) work fastest when the data they are working with resides in memory when the work is being done. Of course, data doesn’t normally sit in memory. It’s stored somewhere, and brought into memory when it’s called by the DB. So, if your database is row-oriented, it’s going to have the best performance if you use it in a way that requires you to operate on a row at time. In our example, that brings one page into memory. (This is just a simple conceptual example, remember. In reality, operating systems bring more than one page at a time into memory. More about that later.)

On the other hand, if your DB is row-based, but you need to get all of the data stored in one column in order to do some operation, then you will have to go access each of the rows, and that means spending time bringing in each row to grab a small piece of data. If you had a column-oriented DB, you could get all of that data more quickly, because it’s all stored together. From the example, all that “2013 Total Order” data is in one column.

But – and this is key – if your DB is column-oriented and you wanted all the data about Alice, you’re going to have to bring in a lot of columns (pages) to get all of that data.

So there is the crux of why there is so much discussion about columnar DBs these days.

Oh, you still don’t quite get it? Well then, let’s go a bit further.

You see, until recently, almost all major databases were row-based. And, for most traditional business processing, this makes perfect sense. Most DB workloads have been what DB experts called OLTP workloads – On-Line Transaction Processing. OLTP workloads are the backbone of most existing business use of DBs. And these applications typically perform best with row-based DBs, because those workloads tend to need many attributes (values stored in many columns) about a single entity. They work with rows. Row-based operations bring in the minimum number of pages from disk when using a row-based DB.

But there are other workloads that perform better if the data in columns can be handled efficiently. Online Analytical Processing (OLAP) workloads frequently want to gather data from columns. For example, if you wanted to know the total of all the values in the column labeled “2013 Total Order” and you had a column-based DB, you would get that by bringing one column into memory and totaling the values, whereas in a simplistic row-based DB, you would need to get data from each row.

Of course, it’s not as simple as all that. Row-based DBs, such as DB2 for i, have added methods to make tasks such as the simple “sum a column” example (or much more complex OLAP analytics) perform well. For example, DB2 for i has couple of constructs called Encoded Vector Indexes (EVIs) and Materialized Query Tables (MQTs) that can be used to great effect. And DB2 for i has ways for customers to block (read many rows at a time) instead of reading them one at a time. In addition to that user-defined method for improving performance, the storage management component of IBM i is smart. Very smart! It has to be; it’s implementing single-level storage. Anyway, because it is so smart, it recognizes that a user is reading rows and then brings data in memory even before it’s requested. This becomes important because many OLTP workloads require a sequential “walk” through the rows, so bringing blocks of rows gets them in memory ahead of when they are needed.

On the other hand, you can’t just add an index to a column-oriented table to get good OLTP performance. As Mark says “It’s a bit like putting Humpty Dumpty together again. The pieces of the row are scattered on many pages of storage. Even if the entire database is in memory, a lot more CPU will be expended pulling all the columns of a row together.”

So, to wrap up this little lesson, the key to deciding which kind of DB to use is to ask yourself what kind of workload is most critical for your DB to support. Though you might want to do both kinds of operations, if your core business needs OLTP, then a row-based DB, with decades of performance optimization, might be the best choice. If your company doesn’t really need fast OLTP, but needs OLAP to be fast, then a columnar DB might be best.

And if you need both, then you will want to look at the mix of techniques available on each. You can decide to go with one, or you might even decide that a combination of the two architectures will suit you best.

Whatever you decide, just make sure that the solution is stable, because businesses really do run on their data.

And that’s it, readers. Database 102. Now, when you read those articles about columnar DBs, you’ll understand why people are talking about them.

Until next time, keep learning.

 

02/10/2014

IBM Knowledge Center Coming Soon for IBM i Documentation

For some of us who have been around for a long time, we distinctly remember the days when systems were delivered with tall stacks of books – reference material for the system and software we delivered. But that was a long time ago. For many years now, that kind of information has been delivered over the Internet, in the Information Center. What you might not have known, though, was that there were actually many Information Centers – more than 800 of them! Though they looked similar, their internals were different, and sometimes those differences showed through. Because of its age, the technology on which the Information Centers are based had some limitations. So it’s time for something better.

IBM will soon be providing a new “one-stop shop” for all your IBM product information needs, including IBM i. This new “Knowledge Center” contains all of the individual IBM Information Center documentation under one system. It’s designed this way to make it much easier to search and find content from any interest area, and to give you the ability to customize your own knowledge space.

All of the Information Center documentation for IBM i releases 6.1 and 7.1 have been migrated to this new framework, and when 7.2 comes out later this year, all of its information will be accessed using Knowledge Center. Eventually the saved and bookmarked links to your favorite pages and content will be redirected to its new home in Knowledge Center.

Great care went into the design of this new Knowledge Center. The designers spent a great deal of time really listening to our users to find out what they struggle with daily when trying to research documentation. This new framework addresses a lot of those struggles, giving you the ability to truly customize and save what is most important to you while also giving you greater capability in finding what is important to you.

It also has a helpful new capability to create your own "Collections". People tend to seek out the same information repeatedly over time. Knowledge Center collections give you the capability to categorize useful content by whatever organization works best for you.

The Knowledge Center is currently in beta, so if you would like to participate, simply use URL:

http://www-01.ibm.com/support/knowledgecenter/

KCimage1

And to view IBM i information, navigate to Software>System Software>IBM i:

KCimage2

Things to note:

  • This beta will run until the end of February 2014
  • IBM wants Feedback! After you have navigated and used Knowledge Center, sign in with your IBM ID and take a few moments to complete the survey on IBM Knowledge Center:

https://www-304.ibm.com/software/support/trial/cst/forms/survey.wss?id=5323

For more information on Knowledge center, visit:

            https://www-304.ibm.com/connections/blogs/techcontent/?lang=en_us

Oh, and for the record, we have already asked that the “KC” people remove the “(AS/400, iSeries and System i)” note from the navigation pane, and we’ve been told those changes will be made – probably even before the Beta ends. There are also some pre-GA links with older names, and those will be changed before everything goes live, too.

I want to thank Kristi Harney in our IBM i development organization for coordinating our use of “KC” and for providing me the information necessary to write this blog entry.

I’ll be sure to tweet when the new Knowledge Center goes live. Expect it some time ahead of the 7.2 announcement.

 

 

 

01/27/2014

2014 and the Modern IBM i

IBM i Logo
Happy 2014, IBM i community. I am having trouble believing the month of January is almost over. Things have been very busy around here already.

We are not ready to announce IBM i 7.2 yet, but the pre-announce activity has been rolling along. Last week I gave a preview of the release to some customers and partners in “Early Ship Program” – customers from India. Then I talked to two different large customers about IBM i Trends & Directions, and they received a short preview as well. (Yes, non-disclosure agreements are in place, in case any IBM executives are reading this. No worries there!)

In my recent customer visits, I have been reminded once again what a “double-edged sword” we have created as we’ve allowed customers to remain on older technology without forcing them to modernize.

I have seen customers that are growing rapidly who are inhibited by database designs that are older than the AS/400. Despite the relational database capabilities that have been a part of this operating system for more than 25 years, they are still running their businesses using older flat-file or hierarchical designs. Now that they are growing, and trying to interoperate with more modern technologies such as Web services, they are hitting limitations that will hamper their ability to handle that growth.

Most of the customers I have seen recently who are encountering this issue are undertaking projects to restructure their data and applications to allow for their present and future needs. Some people like the word “modernization” and some don’t. Whatever you call it, this is something people in IT really need to see as part of the lifecycle of business software. You can ignore it for a while – sometimes for a long while, if you’re using an operating system that continues to support technologies as long as IBM i does – but eventually you have to deal with the changing requirements of your business processes.

That’s why “Modernization” is going to be a big topic for this year. And it’s why we’re about to publish the modernization Redbooks publication. Jon and Susan mentioned it in their blog about Predictions for 2014. Tim talked about it in his iModernization blog last fall. Soon, we’ll be spreading the links around as the book gets published. I have been looking forward to this set of material for a long time, and I am excited that it’s almost “ready for prime-time” because it’s something customers need. We all know that the modern IBM i platform has much more capability than the systems we sold in the ‘90s. We just need to use it that way!