When is an IT Architecture obsolete ?

Ok, perhaps not in a such a drastic way. Here, I am describing an use-case that approves of simple and effective solution that usually renders the resident architectural ceremony obsolete.

Some organizations do suffer from architectural ceremony. Usually arising in a situation when the methodology and process are not tailored for the organization.  Not tailored by kind of business and not tailored by size (complexity) of the business. In which case “priests of the architectural ceremony” are required to translate the not so obvious value of the solutions, to /from the rest of the organization.

Here we have an real life example of a “best” solution, that might be slowed down considerably or even discarded by high priests of the architectural ceremony.

Today, I came over this solution, to solve a requirement for an document/image storage and retrieval. With pretty hefty size requirements at that. By “came over” I mean, after 10+years in the business, I “just” draw it, thought about it (for a day) and then the team (of one) went on to implement it under my supervision. Perhaps one might call this an “architectural stunt” but I beg to differ.

Without false modesty I will admit that it struck me how simple and cost effective this architecture and its implementation is. And still relevant, one year after. And yes, may years after, with a hardware refresh. And this fact is equally if not even more important. This is what customers do value very highly.

See the architecture “sketch” and the specification (image is blurred as it was “saved” from email)

Now just on top of this image any data center technician could very quickly understand how to build this kit and most importantly: what is it for.

A simple but robust and scalable file serving architecture. Or not even an architecture. There is even, enough info on this same image, so that any better asp.net developer, with me present, understands the components specified, and to be developed.

Architecture ( aka “Just a solution”)

HARDWARE SPEC — May 2008

One Load Balancer

Any smaller CISCO router/balancer will do
URL required to be visible to both scanning sites!
(should be something like : http://IR )

Two Servers

HP ProLiant DL320 GP5 (Xeon 3075, 4 GB RAM, 300 GB SATA2)
(in case of overflow, one can add any number of servers, up to the full capacity of the load balancer)

Shared Storage
——————
HP MSA 1000 (fibre channel)
1 year capacity : 4x300GB SATA2 HD
Full 7 year capacity : 20x300GB SATA2 HD drives
(if and when required , inside year 2008, it is easy to extend)

SOFTWARE

Each server:
OS — Windows Server 2003 R2 STANDARD
Web Server: IIS 6.0 (comes with OS)
Run time: .NET 2.0.X

OPERATING REQUIREMENTS

1. developers to be able to connect to IIS remotely and (re)install web sites.
2. user account with admin rights, to be able to have full rights on the local discs and on the Shared Storage too!
3. software using GET and PUT to be able to see them files (images) over a LAN/WAN.
4. HTTP(S)  as the main method of access.

Now. Today , as in July 2009, you and I know most of you will agree with this setup. It can (and it does now, and it will in the future) serve it’s purpose for years to come. And this solution is scalable and resilient and easily incorporated in secure data center.  So the key question is:

Resilience to Change (aka “Why change it?”)

This hardware is very powerful (yes one year latter it still is) and will continue to be at least for next 5 years. An VERY long period in IT time-space. But still why would user/owner of this system, change anything about it, in a near or not so near future? System software for this configuration is stable and good. And solution is made up mostly of just basic OS with basic WIN SVR services used. There is no special third party software here, to be licensed, installed, configured and used. There are few simple ASP pages on top of IIS, and that is all. With source for these pages available for the customer too. There is no UI or client parts. This is pure, robust and simple server side solution that just transparently works.

And this setup is easy to maintain since no special admin or devl skills are needed since there is no special software inside it. And storage backup is a “no wonder” too. It is (after all) just a very large bunch of files and folders.

Architecture Review (aka “Why change this … ever ?”)

And even more importantly: Why not just implement this … I am not advocating this (almost)  “just do it” approach be a replacement for an Architectural Method.

I am advocating the concept by which Architecture works best if the Architectural Method of choice, is first tailored for the business it serves.

PS: I know many companies that would transform this into an simple and thus profitable platform product. For “private clouds” they might claim too.

2 Replies to “When is an IT Architecture obsolete ?”

  1. Sicuramente ci sono sempre spinte ed interessi economici che cercano di farci cambiare i sistemi per poter guadagnarci sopra vendendo hardware e consulenza.
    Spesso i benefici non coprono i reali benefici ottenuti, e sono solo maniferstazione di interesi economici di fornitori e di soggetti interni alla struttura che ottengono vantaggi dal rifacimento di un sistema perfettamente funzionante.

    – La tua soluzione, diventa “obsoleta” nel momento in cui anche un solo pezzo del tuo sistema non risulta sostituibile con uno nuovo uguale trovato in commercio, in quanto rischi il fermo totale della produzione o del servizio.

    – La tua soluzione diventa “critica” se non hai un secondo sistema uguale pronto a subentrare e sostituire il precedente in caso di improvviso fermo. Nella tua soluzione oltre alle macchine devi considerare anche le persone coinvolte e necessarie alla continuazione del processo, in quanto ne fanno parte e lo mantengono.

    Sai tu bene il punto debole del tuo sistema, sulla base di queste considerazioni.
    Per assurdo, il punto debole di questa architettura fra 5 anni potresti essere tu.

    [Google translation]

    For sure there are always pressures and economic interests that seek to change our systems, in order to be able to gain through selling hardware and consulting.
    Often the benefits do not translate into the real benefits obtained, and are only an manifestation of interests of suppliers and individuals within the organization that get benefits from the “renovation” of an otherwise fully working system.

    – Your solution becomes “obsolete” when one single piece of your system is not replaced with a new one as found on the market. Otherwise it risks the closure of all production or service.

    – Your solution becomes “critical” if you have a second equal system, ready to take over and replace the former in case of a sudden stop. In your solution beside the machines you must also consider the people involved and necessary for the continuation of the process, as they are part of it and keep it running.

    Are you aware of the weakness of your system, based on these considerations? Absurdly, the weak point of this architecture in 5 years could be you.

  2. @Andrea : thanks for a comment. I can only decipher it is not a spam .But I do not speak Italian. Can you please re-send in english?
    Also, why hiding behind: a.b@c.d email ?

    –DBJ

Comments are closed.