Hacker News Reader

Comment by derwiki | Hacker News Reader

Pure numbers aren't the important thing here, IMO. How many college students are playing with DB2 or IMS on System Z? It's going to be a very low number, for a few very good reasons:

  * while reliable, System Z is incredibly hard to work with as a programmer or sysadmin

  * in my experience, IBM flat out refuses to give any of their "serious" products to universities. They say "Oh, download DB2 Express-C" and that's the end of the discussion.

If I know there are 1000x more people in the field with Oracle experience on common hardware than DB2 on System Z, I'm going to go with Oracle. Hardware is cheap. Good engineers/sysadmins aren't.

View on HN

There is actually a great number of college students playing with DB2. They usually do so with DB2 Express-C, because it's free and production ready, so it's ideal for students and startups (even though we have large corporations who use it as well). DB2 Express-C is easy to work with, easy to administer, and has the same core code as commercial versions used by the largest companies in the world. System Z provides a nice hardware/software combo, but don't be fooled into thinking that DB2 LUW (Linux/Unix/Windows) is not "serious" or anything but great, really. DB2 provides many useful innovations that are not available to Oracle users. And thanks to DB2 Express-C, many of these benefits are available for free.

DISCLAIMER: I work for IBM in the DB2 team.

This sounds suspiciously like marketing speak. I graduated in 2006, and we never even spoke of DB2 in my time at the University of Michigan.

For curiosities sake, specifically to your post:

What makes it ideal to startups (any cool examples)? What innovations exist in DB2 that are not available elsewhere?

From acangiano's profile:

"I'm a Software Engineer & Technical Evangelist at IBM in Toronto, Canada."

College in Canada has a different meaning than in the U.S.. Here, it means community college. The equivalent term in Canada for an American college would be university. In Canada there are no universities that I'm aware of that have DB2 on their curriculum. There are definitely community colleges that do have it though.

> This sounds suspiciously like marketing speak.

As some of you here may know, I'm a technical evangelist for IBM. I try to remain factual, but I may occasionally be a tad biased. I disclose my (IBM) affiliation in my profile, but edited my comment and added a disclaimer there as well.

> I graduated in 2006, and we never even spoke of DB2 in my time at the University of Michigan.

We have a "DB2 on Campus" initiative (started in July 2006) [1] that has spread to over 250 universities worldwide so far, with about 30,000 students and teachers participating in the program.

> What makes it ideal to startups (any cool examples)? What innovations exist in DB2 that are not available elsewhere?

Unlike different "express" editions provided by other vendors, DB2 Express-C doesn't impose limits on the size of the database, the number of users that can be connected, and so on. This means that a startup can begin using DB2 Express-C for free, while taking advantage of a fast, secure, rock solid, very scalable database that's already being used by large corporations for mission critical work. If you were to hit the jackpot and become highly successful, DB2 would be able to scale with you from your VPS box all the way to a cluster of instances running in parallel within a huge cloud environment.

The free edition will use up to 2 CPU cores, and 2 GB of RAM, regardless of the hardware installed on your server. This is usually plenty, if you consider that many smaller startups don't even have a dedicated database VPS with 2 GB of RAM.

For less than three grand a year, you can also obtain 24/7 technical support from IBM as well as the ability to access further features, like the possibility of setting up a fail over server through the so-called "High Availability and Disaster Recovery" (HADR) feature. This upgrade also enables DB2 to use 4 CPU cores, and up to 4 GB of RAM.

IBM directly implements and supports APIs for various languages, including PHP, Ruby, and more recently Python. When you have an issue with any of those, you have people accountable for, and well paid to solve your problems. If you are wondering, those are open source, too, in case you like doing things DYI style. And we offer advanced tools like IBM Data Studio, free of charge as well. [2]

Learning DB2 is fairly straightforward, and there is a great deal of free information. Amongst which you'll find a course that was extracted from the DB2 on Campus program [3] and free ebooks [4].

Regarding features, DB2 introduced something called pureXML. This allows you to natively store huge quantities of XML data and query it through SQL/XML, XPath, XQuery. It's amazingly fast. [5]

The autonomic features of DB2 are also unmatched, enabling DB2 to self-tune itself, while limiting the amount of work required to administer and optimize DB2.

There are also more "enterprisey" features like DPF, very advanced compression capabilities, pureQuery, InfoSphere Warehouse powered by DB2, and so on. But these are typically of less interest to early stage startups.

1. http://www.ibm.com/software/data/db2/express/students_progra...

2. https://www14.software.ibm.com/webapp/iwm/web/preLogin.do?so...

3. http://www.ibm.com/developerworks/wikis/display/DB2/DB2+on+C...

4. http://www.ibm.com/developerworks/wikis/display/DB2/FREE+Boo...

5. http://antoniocangiano.com/2008/11/04/benchmarking-db2-purex...

I installed DB2 Express-C on my Ubuntu box, and not only did it break itself after one day, it completely broke apt-get. I've never had that experience with any other software on Ubuntu. I tried going through internal channels and no one could point me in the right direction. I doubt that external channels would be anymore responsive.

DB2 deviates from other databases I've used with respect to syntax. Most free information is only from IBM. Compare that to how much information there is about MySQL.

pureXML is not amazingly fast. IIRC, it was orders of magnitude slower than relational queries. I believe development on pureXML was recently canceled.

Occasionally, we run into limitations of MySQL. When that happens, it usually means we're attacking the problem incorrectly. It doesn't mean we need a faster database.

I would be most interested to see examples of startups that use DB2 Express-C; I've not heard of any. When Facebook built one of the largest data systems around with MySQL+memcache and other open source software, it makes you wonder why you need "enterprise" solutions.

EDIT: disclaimer: I used to work as a developer on DB2.

>I believe development on pureXML was recently canceled. Where do you get information like that? Nothing could be further from the truth. >When Facebook built one of the largest data systems around with MySQL+memcache and other open source software, it makes you wonder why you need "enterprise" solutions.

Well, you start with understanding the needs of the application. Facebook and your neighbourhood bank have vastly different needs. When half of your friends see a post on the wall and the others don't for a few minutes nobody gets unhappy or even notices it. Ever wanted to deal with the bank which withdraws money from one account but does not deposit it in to another for a few days? MySQL is a fine database for many applications but just because it is used by Facebook or Yahoo does not make it the right fit for every application. Just because DB2 is used by every fortune 1000 does not make it the right fir for Yahoo either. MySQL is typically not a great fit when you need to scale transactional workloads. Most people don't attempt to run MySQL on anything larger than a 4-processor machine and opt for horizontal scalability based on Master-Slave replication. Can be a solution when slaves falling behind the master is not an issue (e.g Facebook) but will not work very well if you need transactional integrity for your data. Can MySQL be made to scale on larger machines? Yes, if you are Google you can and they will be contributing their bits to v5.4 of MySQL. For the rest of us who need to do enterprise apps it si quite a bit easier to pick a DBMS that does it already and has proven itself. That means DB2, Oracle etc.

DB2 LUW is serious and great (or it was in the mid-late '90s, and a heck of a lot less painful to set up and administer than Oracle), but isn't mainframe DB2 a different code base with a variety of differences and quirks? And how many people know how to do System Z administration or applications programming?

System Z DB2 is written in a completely different language than Express-C. I'm not sure there's even code overlap.

Reply on news.ycombinator.com