|
|
|
August 6th, 2004
05:07 pm - my professional life in Cingular
I think I have figured out my biggest problem, the item which is killing my mood.
I don't like my job right now.
I used to be a developer. I used to program. I have not worked in that capacity for well over half a year, and it does not look like it will change. The group I am in has changed and morphed. Originally I was hired at Bellsouth (Cingular) as a Delphi developer for their point of sale software. When that version of the software was phased out for the new Java based one, the group I was in, disintegrated, with various parts being absorbed into other groups. I went to work for the RTS (Retail Technical Support) branch for a while, developing tools and such. I was also loaned to the "new" POS group to work on weird problems for them, like their reporting engine and other "unimportant" items, that were actually important.
While working on the reporting engine, I wanted a way to "peek" at the internals of the app server while ti was running... to have it report out to me how many requests, how many clients, and how many reports were passing through it. I took this to my then current boss, "M" (his name is impossible to spell... so its an M), and the idea for Hydra was born.
I had done ALLOT of different projects before hydra for Cingular... but it seams that this is the albatross that follows me.
Hydra was both a blessing and a curse. The basic idea behind hydra was to consolidate the miydrid reporting and monitoring tools for all of the applications in the enterprise. All of the shell scripts, all of the SNMP devices, all of the applications that had no monitoring before this, and give the support team a solid, simple view of the actual state of the system. It was a proof of concept. M had no budget for this, so allowed me, and two other developers to take over an office the corner of the building (we have since moved into a new one), and he hid us from the general budget, allowing us to work without oversight from architecture and from the business anylists. A proposal for an application such as hydra had been given, and architecture had torpedoed it, saying that existing testing tools worked, and adding a new layer of complication would not be needed. the support teams could not quiet grasp how this would solve things, so at the time we got no support from them... yet.
So me, and the two developers started programing in a vacume... working on a proof of concept. I had this weird vision in my head of an extensible monitoring and reporting framework, that people could plug in various data sources (heads... ie hydra... the multi headed beast). IT was all abstract and did not make any presumptions on the type or frequency of the data being shuffled around. There were some initial design mistakes (30,000+ sockets.... don't ask)... but the system worked... our proof of concept worked, so we showed it to the support teams, to get feedback, so we knew how the "final" version off the product should have been...
within 30 minutes of showing them the proof of concept in a closed meeting of three "grunts and a minor manager" the vice president of operations was arranging server space on the production floor for hydra. OK... it was not quiet 30 minutes... but you get the idea. From a "we don't need anything like that" to a "mission critical application", the pressure was suddenly on us. Apaerently we pissed of articiture at the same time, since nothing goes onto production floor without an arcticture design document. The architecture department refused to certify it, since it did not go through a standard proposal process, and that the business requirements were unknown, and blah blah blah... basically since they did not give the design, they did not want it. the official was things are supossed to work, business says they want a product, architecture "designs it" (bullshit), and then programing codes to the design specs. To say normal design specs that architecture provides is bullshit is putting it lightly.
anyway... the VP of operations overrides architecture and gets it onto the production floor. Architecture then "freezes" development on hydra (it was at version 1.2c by this time), stating that they are working on a proper enterprise class monitoring application, and that we should not have developed this "in house" and should have gone with a third party product. For the record, I attempted to design hydra around a third party product, but was told no budget meant code it by hand....). Their new solution... "Managed Objects"... which was to replace hydra....
Cool... i did not have any reservations about ditching all of my previous code base (300k+ lines of code)... I was told that hydra was to be pulled off line and managed objects to replace it. They have already paid the money for managed objects, and the final choices made.
a week passed
I asked for a copy of managed objects, so I could get it working in the Load test environment, and to "port" all of the current monitoring assets to it... basically to give RTS all of the current functionality, and what ever it supported... I was going to throw away my entire code base, and embrace managed objects....
more weeks passed.... (hydra 1.2d was deployed in a stealth release... operations gave me the unix password... and turned a blind eye)...
finally... architecture got back o me that managed objects was still in beta, and was not working yet... but they have decided to go with a new product called BMC Patrol... again, don't develop any new hydra... patrol will do everything...
while this was going on, I was working on a new Hydra 2.0... this was a complete rewrite, using everything I learned from the "proof of concept" of the 1.X branch. this was only kept in the load test environment, so it was not production bound (ie it did not break the no new production hydra development rule). Patrol has many faults when trying to monitor an application (15+ minute average response time???!?!?!?!?! hydra has a worse case 45 second response time... normally 8-13 seconds), so architecture commissions a report on how to monitor the Cingular applications with patrol.... sun micro systems consulting branch is contracted to write it... their finding... using Hydra 2.0 and simply use patrol for the display portion... it remove hydra's web interface and use patrol's dashboards... the report is stillborn, architecture cancels the report (still pays) and claims that they will run their own report, that an outside firm could not possibly know how Cingular should do business... their report says that applications should put SNMP agents inside each of the applications... its ignored because of the added cost to each project...
so... while all of this was going on, I was still working on other projects... I was working for the load test group, developing tools and fixing bugs in code when we found leaks and problems. Hydra 2.0 is done... it does 85% of the functionality of the 1.X branch... with only 35k of code... what it does not do is scale across machines (never needed to... wasted code and functionality) and does not send out email or SMS alerts (this is now done with an external perl script)... the API has been cleaned up such that writing a head is stupidly easy, or even reading from JMX, SNMP, RMI, CORBA, Socket, or even a telnet or FTP. So any known application can be monitored... its still born, still sitting on my hard drive....
it was determined that the load test group has no need for programmers, since they should be load testing, not writing tools. Basically in a perfect world, they are right. Each development teams should write their own tools, and all code should be bug free. Basically me and my boss have pissed off arcetecture so much, that they have fought back with the only way they can... politics. So know, I don't have access to the source of any of the projects we load test, and we are fobbiden from writing any new tools. All of the products we used to support (Hydra is dead, GFS given to OPUS... etc) have been moved... and all that is left is running load tests, and writing reports.
I am bored out of my skull. I have tried to transfer to another team... one that does development... but all I get is the run around. I keep hearing "Don't worry... we will take care of you"... or "yea we know, we are working on it"... but nothing happens... when ever I try to get exact dates and times, I keep hearing about "well we need to worry about head count"... or other shit like that. Of the 5 major projects I am personally responsible for... 3 are still in use every day... the code I wrote 4 years ago is still "state of the art" with a rare bug for them to fix (actually I fix it... since nobody looks at code i write because it works... they just bring it to me to fix)... hell... hydra 1.2d which I last worked on in 2002 is finally about to be retired... not because its old and does not work, but because the applications it monitors are about to be phased out... and with the code freeze... there is no new hydra heads to be written...
my boss "leaked" the new hydra 2.0 to a company that we buy tools from, allowing them to see how it works, so now there is a "third party" application that does what hydra does... they did not take code... just concept... and we can finally get arcetecture approval... and they are asking for a broad test across various applications... basically delaying for 6 months while they work on a report which is nothing more than the companies own copy saying why its great. so when hydra goes away... there will be nothing to replace it
and I am still not programing... still not developing... my boss is willing to trade me away to another team... but... nothing happens..
and I want to... I tried applying "anonymously" through the internal hiring portal to someone here... but he hired an incompetent programmer who now does "documentation" only...
while i still don't program...
I am bored, unsatisfied, and my work is suffering because of it...
Hydra I think killed my career here at Cingular... or I don't know... maybe it simply put me into the "omega team" of load testing (My boss also pisses off Architecture on a regular bais). I never wanted to play pilitics. I simply wanted to program a tool that would help people. I love, seriously love, tool development. It makes me the hapiest. I just want to do that again.
|
|
|
