• Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (2)

1 Rodney Rindels commented Permalink

AMEN!

 
I spend hundreds of hours a year teaching the practical aspects of this side of the business to people. Almost every one of those conversations starts with an impossible or unreachable goal because the availability and maturity in the information and context, which typically have quality and maturity that are wanting.
 
You hit the nail on the head with context! Context is the absolute driving force behind reaching a maturation capable of sustaining true service level benefits. I can tell you from my experience in this area, that the one of the most overlooked aspects by most vendors in this space is the service modeling context and enrichment happen way to late in the stack and they loose the opportunity to leverage this for additional grouping and maturation or discarding of false positives early enough, or the data bloat of the model kills performance because of the inability to properly group and aggregate metrics to support the model.
 
Todays routing , redundancy, fail-over, make models so much more complex than previous. Deep Packet Inspection, IPSLA, QoS, etc. Correlating all of this data into context requires some pretty powerful composite type modeling capability. Without the ability to model composite groups of context sensitive data that take into account the true real world aspects; as opposed to the "you shouldn't do things like that anyway" approach some vendors take to their BSM solutions. The reality is , often there are no real choices operations has in terms of acquisitions or growth that can be foreseen or controlled ,so the choice of tools has to supply the flexibility and automation to be useful in the overall approach.
 
The bane of my personal existence is constant model redefinitions because the rate of change typically exceeds the capacity of the operations support teams, and lack of auto discovery to the models because of lack of context. This goes back to having the proper change methodology and approvals which potentially could drive more automation in the discovery and modeling aspects of BSM. Service level modeling in my opinion is still way to "entity" driven to handle all of the Layer 3 and Virtualization of resources happening in today's infrastructures. A more adequate approach overall would be to have a more heuristic type approach to the model that if given enough context can be self remediating where possible. If the industry itself would adopt some level of context agreements and standards and actually stick to them, I think we would be much closer to the goal.
 
How awesome would it be to build rule like this from a single pane of glass!. All nodes matching a pattern of "node%dns%" should be automatically provisioned as dns nodes if they can respond with a dns request, which get the following standard dns metric pack provisioned for performance, and get included in the service models, availabilty, linux nodes, and dns, Setup the inventory schedule to as per the other devices already in the group, collect its node metadata for use in enrichment of alarms and tickets, and lets go ahead and fire off a configuration management request to snapshot the node for configuration management purposes and check it against the gold standard. Add it to the dns maps based on syslocation, and topology views. And shoot while were at it, lets go ahead and open a new entity ticket with the form required to collect the business level metadata we wont be able to auto discover, and once complete inject into our metadata store.
 
My fear for the industry overall is that while IBM continues to make improvements, like composite level performance groupings for composite based thresholding, the reality and big picture is further reaching for the BSM vision, because so much of the vision relies on external data without the context capability, and that context needs to be mediated pre event injection somewhere. Any BSM solution has to have the ability long term to acquire, consume, group, contextualize, aggregate, mediate, enrich, and ease the pain of these external integrations. I keep a pretty close pulse on all the technologies in this space, and I like the direction IBM has been going in for some time for the most part because I think they do understand the value in getting to the goal.
 
I think the passion has much to do with that vision, I hope there are many responses to this blog that I can continue to learn from.
 
Operations is usually a thankless existence, but in my opinion the only place to be that hasn't stagnated and has some challenge left.

2 commented Trackback

[Trackback] [...] Read More: www-304.ibm.com/connections/blogs/netcool/entry/april_12_2010_5_16_pm13?lang=en_us/trackback/ [...]

Add a Comment Add a Comment