process - How should release management be structured for an agile Professional Services department? -
background
professional services departments provide add-on services customers of product.
a lot of these projects small (4-10 hours) , need turned around quickly. additionally, these important projects enhancements customers rely on business.
some challenges are:
- there amount of rework or feature changes customers change mind or make tiny additional request. aside obvious mangement issue (managing scope creep etc.), fact remains there minor tweaks need implemented after project "live".
- sometimes breaks whatever reason , issues need handled expedience. again, these in-production processes customers rely on.
currently, our release management ad hoc:
- engineers manage projects soup nuts, including scoping, customer relationships, code development, production deployment, , project support (for subsequent issues).
- we have dev servers , have production servers. servers exist on-site in server farm. not backed ever, , have no redundancy because not in colo - kind of second class service operations.
- they engineers have full root(linux)/admin(windows) access dev , prod servers. develop on dev servers, , when project ready, deploy prod (basically, copy files up). when issues come up, work directly on servers.
- we use svn source control it's check out dev, work on project, check in necessary, , deploy prod copying files server.
the problem:
the problem number 2 above. servers not treated same reverence operations our product servers (in colo) treated. need servers first class citizens operations. proposal put them in colo, makes them untouchable. if that, need go through operations projects deployed. same arduous , painful process product engineers go through when releasing update our software product.
this take away our agility in responding these tiny projects , issues arise need immediate attention.
the question
how should solve problem? should put servers in colo , work formal release process? how should situation handled?
any making question better welcome!
the servers exist on-site in server farm. not backed ever, , have no redundancy because not in colo - kind of second class service operations.
so want these servers self-serviceable ps engineers, yet have redundancy, backup etc without having go through formal ops processes. can't move them on-site server farm cloud (ec2 or other)? btw, #3 & #4 accidents waiting happen not material main question here.
Comments
Post a Comment