In anticipation of our New York City Summit Event on October 19th, we’re excited to launch our Exhibitor Series, highlighting some of the great partners who will be in attendance. To kickoff the series, we conducted a Q&A with Mike Silvey, Co-founder and Executive Vice President of Moogsoft.
AppD: How have you personally seen the role of IT Operations professionals evolve over the past 10 years?
MS: IT Operations have become more reactive, whereas they were previously proactive. It was originally built around silos of support, each concentrated on a discipline such computing power or databases.
Today, single faults are rarely the root cause of an incident; causality is usually down to multiple faults and changes across different silos of technology either at the same time or cumulatively, leading to performance or capacity degradations which, if left unchecked, will lead to some kind of service impact. Behavior scenarios are changing all time.
The complexity of modern systems has diminished the direct relationships, and now expertise is required across different stacks. Whilst this has created a more “efficient” layering of support (e.g. Level 1, Level 2 etc.), it has diluted expertise, Issues are now much harder for these siloed IT Operations teams to resolve.
The consequence of monetizing alerts, and the resultant billing by alerts, means we look at these critical alerts, not symptoms. It’s a far from satisfactory situation.
AppD: In what ways is the adoption of machine learning (ML) impacting the world of IT Operations?
MS: We’re demonstrating this across our Fortune 2000 customer base: much earlier detection of actionable issues, reducing the mean time to remediate using dynamic teaming and (with predictive knowledge recycling), in some cases indicating the causality before the issue has become an Incident.
The outcomes are all business benefits: no missed SLAs, fewer tickets, fewer support escalations, no all-hands war rooms.
AppD: Can you give me three reasons why Moogsoft’s approach is so unique?
MS: Firstly, we designed Moogsoft AIOps for cloud scale. Secondly, we have not added AI Ops or ML on top of an existing portfolio, but instead identified the key business outcomes needed by IT and built ML from the ground up to help deliver those outcomes. Lastly, Moogsoft ML is real time; it’s not reflective, which needs time to learn experiences and, unable to detect what has not yet been experienced. Moogsoft’s ML discovers outcomes without the need to pre-model or pre-experience behavior, making it agile to change, never missing the unknown unknowns.
We started with a quest to solve the problems suffered by IT Operations and Support today – constant change, the need for agility, silos of support and layers of expertise. Moogsoft is agile to change, makes the silos situation aware of their relationship with each other for a given incident and, by contextualizing incidents and predictive knowledge recycling, reduces the number of escalations
AppD: What’s so clever about your Algorithmic Clustering Engine (ACE)?
MS: ACE is only one of our 16 techniques. ACE is semi-supervised and based on an initial human hypothesis; algorithms will then find these things/evidence that support this. They key thing about ACE is that you can describe what causes impact and leave the algorithms to detect the causality. With Event Correlation systems this would take thousands of ‘brittle’ rules. Brittle because when behavior changes, the rules no longer work.
Machine Learning comes in three forms: Supervised (or learned), Semi-Supervised (or guided) and, Unsupervised…kind of like the Nike version of ML: “Just Do It”.
Moogsoft’s automatic incident detection Machine Learning uses Unsupervised and Semi-Supervised techniques. Moogsoft AIOps does not need to learn, which means that the software delivers outcomes quickly in real-time and are agile. As changes affect the infrastructure, apps and events, Moogsoft adapts automatically. That’s why at Cox Communications, we demonstrated actionable tangible value in less than 10 days while certain competitors did not get out of the starting blocks in 3 months.
AppD: How does Moogsoft reduce that perennial issue, the IT alert storm?
MS: This is one area where [Moogsoft believes that] ONLY Moogsoft can reduce this impact. Moogsoft is is a real time system which uses ML techniques to perform, at machine speed, what humans do: assess textual data. Moogsoft is designed to handle huge burst event volumes (benchmarked at >39,000 per second) and filter noise (by >99.9999%) to surface the incident signal from the noise.
Unsupervised and semi-supervised machine learning techniques along with a linearly scalable architecture ensure that regardless of the source burst event rate, Moogsoft will surface actionable issues from noise.
This enables support operations to be aware of the issue earlier and diagnose and remediate quickly with the right tools such as AppDynamics. That said, often Moogsoft enables operations teams to be aware of issues before they become critical, eliminating the source of the alert storm!
AppD: What role does IT Operations have as DevOps becomes “mainstream”?
MS: The Ops in DevOps needs to become more efficient. You can’t expect a DevOps team to be productive in innovating new software if the whole team is woken 20 times per night with apparent application disrupting issues that they need to investigate…mostly to diagnose that the issue is not caused by bad codeline!
Moogsoft reduces the number of issues that the DevOps team needs to react to, ensuring that only one of the team needs to be notified to be situation aware and, through our contextual information, the DevOps operator can quickly diagnose whether the causality lies with the application or not. In our experience, over 65% of the time, the application is not the issue!
AppD: Why should Summit attendees visit the Moogsoft booth?
MS: If any enterprise has a cloud migration plan, Moogsoft helps makes the transition risk-free with maximum efficiency. In summary, Moogsoft and AppDynamics untether DevOps and Enterprise Application support teams from a reliance on infrastructure Operations, by indicating Application disrupting symptoms earlier, contextualizing the symptoms to enable the DevOps support teams to either use AppDynamics to diagnose and remediate more quickly or, avoid wasting time investigating an issue where the causality is clearly not within the application.
Register here to book your free place at the NYC Summit on October 19th and meet the Moogsoft team there.
Mike Silvey is a 30 year veteran of the ITOM and ITSM industry, having brought SunNet Manager, Remedy and Patrol (before BMC) to market and then, co-founding Micromuse and RiverSoft (IBM Tivoli Netcool and IBM Tivoli ITNM respectively). Mike is an Audi A2 fanatic.