First National Bank of Omaha throws Sun Out

Grifter writes " First National Bank of Omaha said this week that it's nearing completion of a complete changeout of its distributed server infrastructure for a mainframe and blade-server architecture based on Linux. While only 80% complete, the move is already expected to save the company $1.8 million this year in operating expenses and another $9.6 million through 2011." More proof that banks know how to save money.

Comments

Slashdot title is strongly misleading

[marat]

Bank throws out Sun not in favor of Linux, but in favor IBM zSeries mainframe and other IBM's big and small iron. Since words "ibm NEAR cheap" never returned many matches in search (you know the famous "IBM hardware is slow, but expensive"), it's probably an example of some special deal, not a tendency. Nothing to see here... but, probably, "bank managers know how to earn money"?

Re:Slashdot title is strongly misleading

[Old Uncle Bill]

Have you looked and any of the benchmarks lately for Unix based hardware? IBM is the fastest out there and has been for a while. Sun is one of the slowest platforms you can buy, and comparative pricewise to IBM. While Linux on Intel is pretty fast, and definitely the best bang for your buck, most banks are not betting on it yet. There is a lot that goes with that, like a lot of banking software does not run on Linux. Sun is no longer a competitor in the hardware arena, as much as it pains me to say it. Five years ago I was the biggest Sun biggot you could find. Now all of my clients only use Sun to beat up IBM on price, and they typically get a better price with IBM than Sun and much better performance. Don't underestimate IBM performance and pricing, it's not 1987. I price out multi-million dollar infrastructures for my banking clients every day of the week, and trust me, they are looking at cost/performance, not names. The new pSeries server rock in an unbelieveable fashion (and I don't work for them).

Re:Slashdot title is strongly misleading

[Bryan-10021]

Sun has been for awhile trying to turn things around and support x86 hardware from AMD and Intel as well as moving Sparc hardware into the 21st century. Unfortunately Sun is a big ship and it takes time.

Banks are moving to x86 hardware for price/performance benefits and the only UNIX currently available is Linux. So Linux really is the minor reason they are moving. Linux is great but currently has no competition in the x86 server market. I.e. AIX does not run on x86 nor does HP-UX. Sun's Solaris 10 for x86 was only released on February 1st so by next year things may be different.

The main thing is these companies are staying with UNIX, be it Linux or Solaris, AIX ,etc, and not moving to Windows 2003.

Re:Slashdot title is strongly misleading

[Anonymous Coward]

It's amazing which comments are getting modded up in this discussion. I've noticed this before, and it is disturbing which versions of the truth get reinforced at Slashdot. This is definitely not a place for level-headed people, it seems.

Has Slashdot always been so extremely biased? Or has it been hijacked by extremists and moderators with agendas?

Clearly, people have seen Slashdot's PR value in launching FUD wherever and whenever they want. But, clearly, it is not equal-opportunity FUD.

Re:Slashdot title is strongly misleading

[Anonymous Coward]

You're new here, aren't you?

Re:Slashdot title is strongly misleading

[SunFan]

Whatever consensus Slashdot reaches on a given day reflects fashion not reason.

Re:Slashdot title is strongly misleading

[damicatz]

BSDs are available for the x86 platform as well, so Linux isn't the only UNIX-style operating system available. However Linux has the spotlight on it more than the BSDs do.

Suckers!

[HogynCymraeg]

They would have saved even more money using windows. Pah! They need to Get the facts!

Re:Suckers!

[bornholtz]

Actually, they only got rid of 30 Sun boxes. They got rid of 560 Wintel boxes!

Re:Suckers!

[Anonymous Coward]

Now, If Only The Getthefacts Page Was True...

Here's the problem

[zemoo]

Here's the problem - quote:

Kucera said he would have considered Sun Microsystems Inc. products had they been available. But when he began hunting for a way to consolidate his infrastructure in 2003, Sun had nothing to offer in the way of blade servers or Linux.

I'm planning to buy a big Unix Server. Think I can go with Microsoft?

Re:Here's the problem

[SunFan]

"I'm planning to buy a big Unix Server. Think I can go with Microsoft?"

I just had a nano-epiphany: "micro soft" is the exact opposite of "big iron".

Re:Here's the problem

[NateTech]

Small and limp.

Great Day

[Pan T. Hose]

This is a great day for both free software and open source movements. Hopefully Omaha will serve us well as a great example to follow and soon other banks will jump on the bandwagon. I know that I am much more likely to give my money to people who choose their software intelligently and I am sure that I am not alone. GNU/Linux and *BSD variants are certainly the best bets in such an environment. In the name of the Slashdot community, kudos for Omaha! Another question is: what RDBMs are they using for their critical data? Isn't it time for The Industry to finally move from the legacy DB2, Sybase, Oracle and Interbase, to PostgresSQL? This seems like a logical step after converting to Linux. But one step at a time. Let's be patient and hopeful.

Re:Great Day

[Anonymous Coward]

I hope you're just trying to be funny and aren't seriously comfortable with PostgreSQL holding all of your critical financial data.

Re:Great Day

[Anonymous Coward]

And whats wrong with pg IYO?

Re:Great Day

[Anonymous Coward]

> Isn't it time for The Industry to finally move from the legacy DB2, Sybase, Oracle and Interbase, to PostgresSQL? This seems like a logical step after converting to Linux.

I don't know many banks that use Sybase. But a database that chugs along for decades without problems says something about reliability, and Postgres just doesn't have the track record yet. Interbase still wins on the speed factor -- it's just tuned differently.

Anyway, distributed transactions count big in bank apps, and pgsql has no such support. Hot replication, not just master/slave, is also quite necessary.

I like postgresql, but it does not meet all the requirements that Oracle and DB2 do.

Re:Great Day

[Anonymous Coward]

But a database that chugs along for decades without problems says something about reliability, and Postgres just doesn't have the track record yet.

Are you kidding? From Wikipedia.org/Postgres:

PostgreSQL has had a lengthy evolution, starting with the Ingres project at UC Berkeley. The project lead, Michael Stonebraker had left Berkeley to commercialize Ingres in 1982, but eventually returned to academia. After returning to Berkeley in 1985, Stonebraker started a post-Ingres project to address the problems with contemporary database systems that had become increasingly clear during the early 1980s. The code bases of Postgres and Ingres started (and remain) completely separated. The resulting project, named Postgres, aimed at introducing the minimum number of features needed to add complete support for types. These features included the ability to define types, but also the ability to fully describe relationships - up until this time widely used but maintained entirely by the user. In Postgres the database "understood" relationships, and could retrieve information in related tables in a natural way using rules. Starting in 1986 the team released a number of papers describing the basis of the system, and by 1988 the project had a prototype version up and running. The team released version 1 to a small number of users in June 1989, followed by Version 2 with a re-written rules system in June 1990. 1991's Version 3 re-wrote the rules system again, but also added support for multiple storage managers and for an improved query engine. By 1993 a huge number of users existed and began to overwhelm the project with requests for support and features. After releasing a Version 4 -- primarily as a cleanup -- the project ended. Although the Postgres project had officially ended, the BSD license (under which Berkeley had released Postgres) enabled Open Source developers to obtain copies and to develop the system further. In 1994 two UC Berkeley graduate students, Andrew Yu and Jolly Chen, added a SQL language interpreter to replace the earlier Ingres-based QUEL system, creating Postgres95. The code was subsequently released to the web to find its own way in the world. 1996 saw a re-naming of the project: in order to reflect the database's new SQL query language, Postgres95 became PostgreSQL. The first PostgreSQL release formed version 6.0. Subsequently a group of database developers and volunteers from around the world, coordinated via the Internet, have maintained the software. Since version 6.0, many subsequent releases have appeared, and many improvements have occurred in the system; on January 19, 2005 version 8.0 became the current release. Although the license allowed for the commercialization of Postgres, the Postgres code did not develop commercially with the same rapidity as Ingres -- somewhat surprisingly considering the advantages Postgres offered. The main offshoot originated when Paula Hawthorn (an original Ingres team member who moved from Ingres) and Michael Stonebraker formed Illustra Information Technologies to commercialize Postgres.

[emphasis added]

Your point again?

Re:Great Day

[Anonymous Coward]

Anyway, distributed transactions count big in bank apps, and pgsql has no such support. Hot replication, not just master/slave, is also quite necessary.

The informations you are looking for are here and here. See also this.

Why can't people see?

[doc modulo]

One day, Free and Open source(TM) software is going to be 80% of software out there in the world (give or take a little).

I think Sun would be wise to concentrate on hardware and selling services, like IBM.

Re:Why can't people see?

[SunFan]

I think Sun would be wise to concentrate on hardware and selling services, like IBM.

I guess that's why their new pricing model for Solaris 10 is based on support and service _only_. Sun have clearly shown they can adjust their business model to stay afloat. There is no price advantage, now, of Linux over Solaris 10--customers will choose based on their wants and needs, not price. The only reason, probably, that this bank could not stay with Sun is that they started planning a long time ago before Sun's new pricing.

AFAIK

[hummassa]

WRT: "I hope you're just trying to be funny and aren't seriously comfortable with PostgreSQL holding all of your critical financial data."

PostgreSQL is a nice, security- and data-integrity-oriented DBMS, with a good and proven track record, and I see no problem with it holding my financial data.

As a matter of fact, I really prefer it over proprietary DBMSs where I have no guarantee of what is going on.

And WRT the data itself, I think banks know better -- and therefore are very rigid -- about backup procedures, etc.

Nothing to see here, IMHO.

Re:AFAIK

[DrSkwid]

I'd prefer they just stored it in plain text over proprietry binary formats, no matter how open they are.

Re:AFAIK

[Anonymous Coward]

I'd prefer they just stored it in plain text over proprietry binary formats, no matter how open they are.

If you use Perl then there are DBI modules (DBD::AnyData, DBD::CSV) to have a SQL frontend to plain text files (fixed length formats, variable width comma- (CSV), tab-, pipe-, colon- or anything-separated values, paragraph records, XML...). It's easy to use, and even quite fast. Great for writing simple apps that one day might need more traffic/data. Need more throughput? Just change the connect to use DBD::SQLite. Want a bigger RDBMS? Just change it to use MySQL or PostgreSQL. etc. When you use DBD::AnyData for simple programs with small data sets, you have your data in text format great for processing with grep/sed/awk textutils or even manually with your $EDITOR of choice (vi or Emacs) and yet scalable to giant data sets on Oracle with only one line of your app to change if you write it correctly in the first place. Great stuff.

Proof?

[FullMetalAlchemist]

No Mr. Cowboy, you can't prove anything _in the future_, only lates verify that your predictions where true.

In essence, I doubt it will save or cost anything. Actually, it might intruduce additional costs, as such changes often introduce hidden and unplaned "soft" changes.

Anyway, is this news worthy?

If they had done this with BSD

[MarkRose]

If they had done this with BSD, you know there would have been a devil in the details.

Other banks?

[thule]

I wonder how many other banks are doing this, but don't shout it from the rooftops. My understanding is that Washington Mutual has a Linux team. It would be interesting to find out how extensively they are using or planning to use Linux.

I know they recently deployed a huge web farm of Windows boxes for some business web site. It's not surprising they use Solaris, AIX, Windows, Netware, and zOS, but how much do they and other banks use Linux?

Anyone have first hand knowledge that they can share?

Re:Other banks?

[dpilot]

I've gotten quite a bit of email from Washington Mutual, lately. They want me to verify my account at their web site.

I'd really love to help them out, but I don't have a Washington Mutual account. Maybe I should direct them to that guy in Nigeria.

Re:Other banks?

[Old Uncle Bill]

I wonder how many other banks are doing this, but don't shout it from the rooftops?

Speaking from direct experience, not that much. They are starting to look into it seriously, and I say go for it, but they are starting really small. I suspect in the next few years many of their mission critical apps will be running on linux, but not many of them are today. Oracle has really been pushing RAC/10g on linux, with mixed results. A few of my clients have gone to marketing seminars with Oracle and come out preaching, but once their dbas go through the actual classes they have completely bailed on the idea. I don't think that speaks negatively on linux, just the whole RAC solution in general. It's a big bet and the reasoning is simpler than you would think. It doesn't come down to whether the solution actually works. It comes down to not wanting to be the infrastructure manager when the press release goes out saying your RAC/Linux database went down for several hours in the middle of the day. No one wants that Computerworld interview. I'm a huge fan of linux, and I think it is ready for prime time, but don't look to financial institutions to set the pace. They are in the business of making money, not trying to be bleeding edge on technology. The savings honestly is not that great in the scheme of things. Consider saving a couple hundred grand on going with a linux solution with an hour of downtime that costs 500K. This does not mean linux is less reliable, but the suits are going to question your motives for going with this "linooks" thing.

FNBO

[two_socks]

I worked at FNBO for a number of years - the thing that impressed me most about them was the consistent high quality of everything IT there.

I went through the rollout of a few software packages, and they always "just worked" right out of the gate. The uptime on all of the systems was just as impressive.

The Good, the Bad & the Ugly

[Dark Coder]

[cue Ennio Morricone spaghetti western music....]

While Angel Eyes Gates laggerly leers on....

"They were late to the game,"...

Broken-heartedly, Tuco McNealy rides off to the sunset...

...Sun had nothing to offer in the way of blade servers or Linux.

Gunsmoke hangs in the air around the penguin with No Name.

Re:The Good, the Bad & the Ugly

[brsmith4]

God, I hope you work for an advertising firm so we can see brilliance like that on television. Bravo!

Re:The Good, the Bad & the Ugly

[Anonymous Coward]

Mark him as your Slashdot friend?

Windows Errors

[kereira]

You know most services run on Windows. I hope some more companies can follow from this example. I have been around in museums with small interactive computers with windows errors popped up! It's insane. If they just used Linux it would be:

a) Cheaper

b) More reliable

Good on 'em.

Re:Windows Errors

[Anonymous Coward]

Linux is not cheaper if you have to hire a new, qualified IT guy instead of the stupid MSCE. What museum has *nix gurus lying around?

They would do better to buy Macs ... rock solid BSD core, pretty display, stays up for months, Applecare warranty is the shit...

Convert, but have to fight the sun engineers

[Bruha]

Where I work the datacenter is primarily sun based hardware. Sure we have a few multiproc machines running redhat and I cringe every time I see a department say were going to use a sun workstation for each employee. When in fact for each complete workstation we could have better performance from a properly configured Linux machine and get 3 of them for the same price.

When you get into the lets buy Linux servers the sun engineers are quick to tout that the sun servers are better and every other excuse in the world. There are no official "Linux Engineers" in the company so our counter arguements are always brushed off like we dont know anything.

Re:Convert, but have to fight the sun engineers

[SunFan]

Tell your Sun engineers that Sun sells Linux, too, on really good hardware. These articles touting migrations to/from Sun are annoying as they could just as well be migrations from Sun to Sun.

Re:Convert, but have to fight the sun engineers

[superpulpsicle]

Sun fanboys have really moved up the list by the end of 2004.

Most annoying fan boys
----------------------
1.) Apple
2.) M$
3.) Sun
4.) EMC
5.) IBM ...

Re:Convert, but have to fight the sun engineers

[nathanh]

When you get into the lets buy Linux servers the sun engineers are quick to tout that the sun servers are better and every other excuse in the world.

The Sun servers are pretty good.

• Hotswap PSU, hotswap Ultra320, hotswap CPU, hotswap blades, hotswap memory, hotswap PCI-X.
• LOM standard on every device, ethernet and serial varieties, remote poweron and off.
• Binary compatibility from a single-CPU Netra all the way up to a 100+ CPU mainframe.
• Automatic detection and disabling (without downtime) of faulty CPU, DIMM, etc.
• Dynamic domains on the big iron, including partitioning of CPU and memory.
• Extensive capacities for hardware expansion from even the cheapest chassis.
• Enterprise management software; monitor 1000s of servers for hardware failures, capacity limits, etc.

Sure, none of that says desktop, but in big computing centres those Sun server features are pretty damn useful.

There are no official "Linux Engineers" in the company so our counter arguements are always brushed off like we dont know anything.

Well, maybe you're thinking about the problem in a different way. You're thinking about the software. They're thinking about the hardware. It isn't the case that you don't know anything, but rather that you're focussed on problems that they don't think are important.

For what it's worth, I don't work for Sun. I think Linux is superb. I've been running Linux exclusively as my desktop for 13 years now. I even run Linux on this Apple PowerBook; no dualboot, just Linux. But I'm not so blinded as to ignore Sun servers and Solaris. They're both good too. There are some situations where I would never recommend Linux.

Re:Convert, but have to fight the sun engineers

[Anonymous Coward]

Hmmm...you left out Linux fanboys and GPL fanboys, the most annoying of all.

People....saving costs, not Hardware really !!!

[newnovice]

Its the people that they are getting rid of is saving them money.....not really the UBER EXPENSIVE IBM Z series Blades....IBM is still expensive... They are reducing their SUN System engineers from 24 to 8, from what i read.... i guess no more job market for such OS specific experts out there !