As I am closing out 2015 and starting into the New Year, I want to give you a few sound bites on what I have seen the last 12 months. I am very interested to hear from you if you agree to these observations, and/or interpret these symptoms differently …
Revitalizing Open Source – upside down
We have been seeing Open Source in ITSM in pockets. The biggest success certainly was the monitoring tool Nagios, another area is provisioning through Chef and Puppet and basic SNMP monitoring with Cacti or PRTG. Most of the other ITSM capabilities – Event Management, Scheduling, Helpdesk, Automation – were not that approached by Open Source. Therefore, we have considered the strategy where agents (the manage-to side) become commodity, and we would focus on Infrastructure (the manage-from side).
However in 2015 I have seen this actually the other way round. I have seen clients that build infrastructure using open source: Kafka, Spark, Hadoop, Cassandra, Solr and use commercial off the shelf agents to feed into this infrastructure. I have engaged with two clients (to financial services clients, one I the US, and on China), that built their own infrastructure by integrating various open source technologies together. So, the infrastructure becomes commodity and the agents is where clients need vendor support.
Standards evolve, for instance ELK (Elastic, LogStash, Kibana) or SLK (LogStash, Solr and Banana). Even within IBM, we are increasingly adopting open source components for our own products: APM uses Kafka and Cassandra; IOA-LA has adopted the SLK stack; etc. It’s a variation of the make-vs-buy discussion: what technology do we have to develop and maintain within IBM, and what components can we acquire through other sourcing options, open-source being one of the options available.
My takeaway for 2016: Start majoring on Open Source technologies, as they are real and becoming increasingly relevant to ITSM.
We have been preaching “do more with less” for many years. Our products have added and enhanced capabilities to support this target: baselining, topology-based correlation, automation, rule-based processes, etc. etc. At the same time, the environments we are managing become increasingly complex and dynamic – driven by topics like Agile, Virtualization, DevOps, CAMSS. It now feels that the traditional approaches hit a glass ceiling and will not be able to keep up with the pace. If feels like Herakles fight against Hydra.
We need new approaches to win this fight. Cognitive operations could be one of these approaches. Rather than coding rules, the system will do. Rather than instructing the system, the machine will learn by itself.
I am currently engaged in an exciting project where we apply Watson capabilities to ITSM. We digest incident, problem, and change records as well as chat transcripts. Using National Language Processing, we make sense of this non-structured data. We are also looking at structured data like CMDB information (Cis and their relationships) or events. By stitching this information together, we hope to be able to identify new knowledge and make this knowledge available: Is a new incident similar to one we already received? What has been done back them to restore the service? Who was the SME that could solve the incident?
We are also using Watson technologies in IBM Support. Several activities exist – Watson for TSS, Watson in the Middle, Watson for IT – all aim to support (or relieve) the SME from doing mundane tasks.
My takeaway for 2016: Cognitive Operations can be the unique value proposition for IBM. We have a lot of technologies at hand, and we need to apply them to ITSM to move the needle forward. And to stay relevant to our clients
LoB vs. IT: stay behind or lead
We’ve been also talking a lot that the buying behavior of our clients is changing. Instead of the IT organization, new projects are increasingly generated from the line of business. This is certainly true for many of the middleware functions: Decision management, Integration, Mobility, Analytics, etc. IT Service Management typically is a function of the IT organization, and the other departments are only interested in a stable and performing IT service.
Well, yes and no. There are the typical operational activities that the IT department takes care of: 24×7 operation with the need for availability monitoring and event management; scheduling of 1000s of jobs at the right time in the right sequence; a helpdesk. But then there are functions that blur the line towards the LoB. Applications are not just up or down, they may become slow. In order to know why, and even more important in order to know what do to fix it, deep application skill is needed – and this skill resides in the LoB.
I had some interesting meetings in 2015. One was a European Insurance. Their IT department lost relevance, the LoB / application department basically took back control. Over 200 small departments develop their applications, and select their individual service management tool strategy. Their interface to IT was minimal, they only generated availability events for 24×7 operations. The rest was done by the (200) devops teams themselves. The new CIO declared the mission “coding is cool” and expects every (!) employee to write at least 2000 lines of code each year. In this evolution, the IT team was left behind.
Another interesting meeting was with a large payment processing firm in the USA. We discussed Analytics and the role of Data Scientists, quickly confirming each other that the key skill for the data scientist is not the analytical / mathematical skills, but rather the domain know-how to ask the right questions, distill the data, and then conclude from the findings. We also brainstormed what one could do with all the data that systems management has: If you understand the performance and capacity data, you could proactively inform a branch that they are getting less traffic / revenue / visitors compared to a typical day. With other information they may know why, and they see it as the data is coming, not just at the end of the business day. So, IT could step up and provide data -> information -> insight to its users in much different ways than before.
My takeaway for 2016: LoB will continue to become more relevant. Rather than applying basically the same tools / functions to different users, we need to think what insight we can derive on top. Service Management can act as the integration point amongst many different aspects (System, Network, Application, Security, Assets), so let’s identify what relevance we can offer to these new stake holders.
Modern ways of Application Development – I do it myself
It certainly feels that people code differently than just 5 years ago. People get on a SaaS-based development platform, instantiate a database and start coding. If they don’t like the SQL database, they swap for a No-SQL database, and tomorrow they may try a Graph-DB. Applications are no longer written using a single programming language. I used some code from GitHub, look for sample code on Slashdot and compile them together – btw compile not in the definition of a programming language, many languages are back to interpreter mode. You have seen how easy it is to develop on BlueMix. You have seen us organizing Hackathons for clients or at Universities. This is real.
Now in the past, people leveraged the skill in the organization – and the service of the organization. If they need a database, they went to the DBA and requested an instance. Today they do it themselves. Because the technology allows them to do this.
And then there are new topics coming up all the time: APIs, Micro-Services, Docker, etc.
My takeaway for 2016: This is a skill we need to start learning: instead of always asking an SME, we need to jump on a system and try it out ourselves. Install a GraphDB, visualize data with D3, scale out using Softlayer or AWS… How are applications developed, how disruptive are DevOps, BlueMix, API-economy… It is more important to get out of the comfort zone, than to always achieve the optimal path forward.
Run books run, and not limp
I have been involved in Runbook Automation for quite some time now. I am a big believer in automation, and despite all the changes and disruption, there is still quite some fat that can be reduced through automation. It was shocking to observe that many enterprise clients still get a lot of events because a filesystem or tablespace became full, a process or service died, a job abended.
I was happy to engage with the development team to build a new offering for runbook automation. It was exciting to collaborate and build a new offering from scratch. And I truly believe it is a shiny baby when it come out.
My takeaway for 2016: we talk a lot about Design Thinking and Minimum Viable Product (MVP). There are a lot of roadblocks in developing and implementing runbook automation. Time is there. Lets do it. Now.
There are probably a few more thoughts in closing 2015: The role of architects to translate the vision into actions that can be implemented by SMEs. Or Will 2016 be the year of Service Management as a Service (SaaS)? .Or. an emerging trend if IT4IT to amend ITIL and make it more actionable. Food for thought, maybe for the next blog post.
What are your takeaways for 2016. I’d be interested in your point of view …