Technology Marketing Corp. next month debuts The New Intelligence, the umbrella brand for a collection of events and exhibits focused on how artifical intelligence and machine learning are impacting business, communications, and networking. Sameer Padhye, FixStream’s founder and CEO, will be giving The New Intelligence keynote. During that presentation he will discuss how AIOps can enable self-healing hybrid IT infrastructures. (And FixStream’s CTO will speak at AIOps Expo, part of The New Intelligence)
FixStream is an AIOps solution provider. What is AIOps, and why are we hearing about it now?
Padhye: AIOps, artificial intelligence for IT operations, is a strategic segment defined by analysts such as Gartner (News - Alert), Forrester, EMA, etc. Gartner, for instance, says AIOps ‘platforms are software systems that combine big data and AI or machine learning functionality to enhance and partially replace a broad range of IT operations processes and tasks, including availability and performance monitoring, event correlation and analysis, IT service management, and automation.’
In simple terms, AIOps is what’s required to correlate the millions of data points that multiple monitoring tools generate across IT, allowing IT operations to predict when the next outage will occur.
Could you provide an example of how one of FixStream’s actual customers is benefitting from your AIOps solution?
Padhye: Many customers have successfully deployed FixStream AIOps solution in production environments.
For instance, Broadcom has deployed it to increase the uptime of their mission critical production line. Maxim Integrated leveraged the autodiscovery capabilities of AIOps to accelerate data center migration, while City of Las Vegas deployed it to predict outages in their Oracle ERP-based processes to increase cash flow.
Who’s the typical buyer of FixStream’s AIOps solution?
Padhye: The typical buyer is a VP/director of IT operations in an enterprise that is struggling with one or more of the following pain points:
• Unplanned downtime of mission-critical applications;
• High MTTR/MTTD costing [a] significant amount of resources and potential revenue/CSAT loss;
• Complex/risky migration to cloud and/or data center migration/consolidation;
• Inefficient IT assets utilization; and
• Costly/inefficient audit/compliance.
How does your AIOps solution work with a company’s existing network elements?
Padhye: FixStream AIOps platform interfaces directly to the existing infrastructure through both north- and southbound interfaces. It accesses each of the domain (compute, network, storage, etc.) via service accounts (through technologies like SNMP, SSH, vendor-specific APIs, cloud subscription information, etc.) as well as via connectors to third-party tools (like APM, ITSM, etc.).
The FixStream platform also provides a rich set of APIs to facilitate the bi-directional integration with existing applications like a CMDB, orchestration/remediation solutions, etc. This open approach enables seamless integration while preserving the IT investment already made into existing solutions.
Does a company have to use a particular supplier’s firewalls, routers, switches, or other gear if they want to use FixStream for AIOps?
Padhye: The FixStream platform supports over 250 technologies, vendors, products across IT’s data centers, including all the major vendors (Cisco, Juniper, Oracle, SAP, Microsoft (News - Alert), AWS, Azure, etc.).
A detailed list of supported technologies is represented in the datasheet. FixStream also supports custom applications (for instance built on Java).
Your company recently announced that AIOps is taking Oracle’s ERP platform to a whole new level. What does that mean?
Padhye: Oracle ERP is one of the most critical applications that a company requires, in that it supports mission-critical business processes like order-to-cash, procure-to-pay, supply chain management, etc. Should any of these processes fail, the company could suffer major setbacks. They could miss their earnings, they could lose millions of dollars in revenue, and it could severely impact their brand and customer reputation.
The owners of such applications, though, don’t have the end-to-end visibility necessary to detect or predict what could make these app fail. What happens if a network link goes down? Does that impact the O2C process? Can that slow down my orders at the end of the quarter? How long will it take to find out?
Traditional domain-centric tools provide a silo view of the customer IT environment and cannot correlate in real time the millions of data points across the entire stack. FixStream AIOps platform is the industry-first solution that can not only detect issues across business transactions, application services, and infrastructure, but it can actually predict when the next outage of an Oracle ERP process will occur. By the way, this can be done for a broad range of Oracle applications (including Oracle EBS, Oracle Siebel CRM, Oracle JD Edwards, etc.), SAP (News - Alert), and custom apps.
The AI part of AIOps stands for artificial intelligence. How does AI come into play in the FixStream solution?
Padhye: AI comes into play in multiple forms.
Dynamic thresholding and multivariate anomaly detection detect[s] anomalies off a normal or expected behavior. It reduces noise, and accelerates root cause analysis across any variable (CPU, memory, etc.) supporting a specific application.
Sequential pattern analysis and incident prediction detects a pattern of events across the entire stack – from infrastructure to apps to business transactions – so that it can predict when the next outage will occur.
[AIOps also enables] disk capacity predictive analytics to proactively plan to add more capacity before it negatively impacts the business. Using the weighted slope one algorithm it analyzes the historic trend of utilization of an infrastructure resource and predicts when the infrastructure entity will run out of capacity across business transactions and underlying application and infrastructure entities. Representation of time-series metrics of the impacted vectors in context with a business application or device group … allow[s] a user to visually go back and forth in time and examine the behavior of multiple metrics in time series which are a part of group of anomaly events.
[It can] identify a sequence or group of anomaly events for transaction, application, and infrastructure entities that may impact a business application. For example, unplanned events such as DDOS attack[s] and planned business events such as Black Friday (News - Alert) or promotional events trigger [a] group of anomaly events across business transactions and underlying application and infrastructure entities.
Some people say artificial intelligence isn’t ready for prime time and that what some suppliers are calling AI isn’t really AI at all. What are your thoughts on that?
Padhye: There is a lot of truth to it. AI, for instance, will not help out if it is based off very inaccurate data sources. If the IT inventory, for instance, is not refreshed in near-real time, or is not 100 percent accurate, the ML models will generate incorrect predictions. That is why we recommend all accounts to first start with an agentless autodiscovery process that can be synchronized with their CMDB. This will create an accurate single source of truth that will increase the efficacy of any subsequent service management operation.
Likewise, AI cannot help out much if you cannot start with a correlated set of data points. What we recommend is to develop a contextual, applications-centric view of the environment upon which to correlate events. Without it, you will simply reduce noise but not actually solve the issue.