(Editor\u2019s note: Enterprise Management Associates took a look at how individual organizations choose, implement and use network performance management (NPM) tools. In this article, EMA\u2019s research director for network management, Shamus McGillicuddy, presents some findings and suggests some best practices based on EMA\u2019s recent report \u201cNetwork Performance Management for Today\u2019s Digital Enterprise\u201d that\u2019s based on a survey of 250 network managers.)\nThe typical IT organization has three to six network performance management (NPM) tools installed today, and if they remain siloed, network operations will be fragmented and inefficient \u2013 a persistent challenge for network managers for many years.\n\nEMA asked 250 network managers to identify their preferred procurement strategy for NPM tools and found that enterprises have a strong preference for a fully integrated, multi-function platform. However, they rarely achieve this state. For example, EMA found that enterprises that currently have 11 or more NPM tools are the most likely to state a preference for this fully-integrated strategy. Thus, while they profess a desire to consolidate tools, they are not succeeding.\nWhy so many NPM tools?\nPart of the problem is that enterprises collect and analyze so many different types of data with NPM tools. Infrastructure metrics collected via SNMP MIBs and traps are a foundational data source for NPM, but they don\u2019t contain all the answers a network manager needs from NPM tools.\nMost enterprises also collect traffic data, including network flows, packets, or both. EMA research also observed strong interest in synthetic traffic generated by active monitoring tools. The most popular source of data for NPM analysis is management-system APIs. In other words, network managers have strong interest in pulling data from other IT management systems into NPM tools for contextual analysis.\nGiven this data diversity, tool fragmentation is inevitable. After all, no vendor excels at collecting and analyzing every class of data mentioned above. They usually excel at one or two classes of data types, meaning that enterprises inevitably acquire additional NPM tools to cover gaps in visibility.\nCorrelation across NPM tools\nEMA asked survey participants to reveal how they correlate insights across multiple NPM tools. The most popular approach (25% of respondents) was the use of a network operations management platform or manager of managers that pulls insights from multiple NPM tools. These platforms are typically good at event management and alarm correlation across multiple NPM sources.\nNext, 19% cited direct integration between point tools so that one tool can correlate insights pulled from another. This approach can get complicated if an enterprise is using more than two tools.\nAnother 19 percent integrate their NPM tools with an artificial intelligence for IT operations (AIOps) advanced IT analytics platform, 15% integrate NPM tools with a service management platform, and 14% stream NPM data to a data lake for correlative analysis. Only 7% claimed to perform these correlations manually, which is good because it\u2019s an inefficient and error-prone technique. A handful claimed that they perform no correlation across tools.\nEMA also asked enterprises how successful they were within this cross-tool correlation. Twenty-seven percent said they were very successful, and 49% were successful. The rest were somewhat successful, somewhat unsuccessful, or uncertain. EMA classified this last 24% as \u201cless successful.\u201d This question about success allowed EMA to search for potential best practices.\nEnterprises that manually correlated insights across tools tended to fall within the \u201cless successful\u201d cohort. The most popular approaches to correlation \u2013 direct integration between tools and integration with a manager of managers \u2013 had no statistically significant associations with success.\nBest practices\nHowever, three less popular approaches to cross-tool correlation were preferred by successful organizations.\nSuccessful enterprises preferred Integration with a service management platform or correlation via streaming of NPM data into a data lake for analysis. Very successful enterprises integrated their NPM tools with an AIOps platform.\nEMA believes that these three latter approaches are potential best practices for addressing the problem of NPM tool sprawl. AIOps tools appear to be the best option. There are many standalone AIOps platforms that can fulfill this need, such a Moogsoft and Splunk. Furthermore, some NPM vendors, such as Broadcom (formerly CA) are developing their own AIOps platforms that can correlate insights across their suites of NPM and IT operations management tools.\nEMA recommends that enterprises investigate AIOps platforms if they are struggling with network management tool sprawl. However, integration with service management platforms or the use of a data lake with a data analytics stack may also prove helpful.