As you may have heard, in early March, Hewlett Packard Enterprise announced its intention to acquire the independent manufacturer of hybrid and flash arrays, Nimble. On April 17, this purchase was completed, and now this company is 100% owned by HPE. In countries where Nimble was previously available, Nimble products are already accessible through the Hewlett Packard Enterprise channel. In our country, this process will take longer, but it is expected that by November, Nimble arrays will occupy their niche between the higher configurations of MSA and 3PAR 8200.
Along with the integration of manufacturing and sales channels, HPE faces another challenge: leveraging the software developments of Nimble InfoSight, which go beyond just storage systems. According to , InfoSight is a leading platform in the industry for predictive IT analysis, the benefits of which other manufacturers attempt to replicate. HPE currently has an equivalent— , however, both IDC and Gartner rated the Nimble product significantly higher in their 2016 'Magic Quadrant for All-Flash Arrays.' What are the differences?
InfoSight changes the approach to managing infrastructure related to storage systems. It can be quite challenging to pinpoint the source of issues that may arise in the 'virtual machine – server – storage system' link, especially when all these products are supported by different manufacturers (I remind you that in HPE's case, ). It would be much easier for users if comprehensive infrastructure health analysis was performed automatically at all levels of IT through which business application transactions pass, and the results were presented as a ready-made solution. Ideally, this should happen even before the problem arises. Nimble InfoSight software does just that, enabling unique results: data availability Essentially, on entry-level systems, automated prediction of potential issues (including those beyond storage systems) allows for preventive measures in 86% of cases. Without the need for a system administrator or calls to support! In general, if you want to spend less time managing your information system, I recommend taking a closer look at InfoSight.
How is it done?
One of the key distinctions of the NimbleOS operating system is the greater amount of diagnostic data available for analysis. Instead of standard logs and system health metrics, a vast amount of additional information is collected. Developers refer to the diagnostic code as "sensors," which are embedded in every module of the operating system. The installed Nimble base has over 10,000 customers, and tens of thousands of systems are connected to the cloud, which currently holds 300 trillion array metrics accumulated over years of operation, analyzing millions of events every second.
When you have such a large amount of statistical data, all that's left is to analyze it.
It turns out that more than half of the problems that cause slowdowns in business application I/O operations , and other manufacturers focusing solely on storage systems simply cannot adequately understand the service case in most instances. By correlating array data with other diagnostic information, one can identify the true source of issues all the way from virtual machines to the array disks. Here are some examples:
1. Performance diagnostics – is quite a complex task for intricate IT infrastructures. Analyzing log files and metrics at each level of the system can be time-consuming. InfoSight, based on the correlation of multiple metrics, can determine where the slowdown is occurring - on the server, in the data transmission network, or in the storage system. The issue might be with an adjacent virtual machine, possibly a misconfiguration of network hardware, or an optimization of server settings may be required.
2. Invisible problems. A specific sequence of metrics forms a signature that allows predicting how the system will behave in the future. More than 800 signatures are monitored by InfoSight in real-time, enabling the detection of issues beyond the array. For example, one customer experienced a tenfold performance drop after upgrading the storage system's operating system, related to the hypervisor's operational characteristics. Based on this case, a patch was not only released but also automatically prevented similar situations in another 600 storage systems, as the signature was immediately added to the InfoSight cloud.
Artificial Intelligence
This may be too bold a statement to describe the operation of InfoSight, yet advanced statistical algorithms and predictions based on them are the key advantage of the platform. The algorithms used by the platform include autoregressive forecasting models and Monte Carlo simulation, which help predict 'random' events that may seem arbitrary at first glance.
Data on the current state of the infrastructure allows for precise sizing for the upgrade of information systems. Since the deployment of new components, InfoSight gathers data for subsequent analysis, making the mathematical model even more accurate.
The platform constantly learns from the installed base that has been created by customers over the years of Nimble's existence, facilitating simpler and clearer support tasks for systems now under Hewlett Packard Enterprise. The number of 3PAR arrays currently operational among customers significantly exceeds the respective Nimble benchmarks. Consequently, InfoSight's support for 3PAR will provide an even more comprehensive picture for statistical analysis of IT infrastructure metrics. Undoubtedly, modifications in the 3PAR OS will be needed, but on the other hand, not everything integrated into InfoSight is unique to this platform. Therefore, we look forward to news from the combined development team of Hewlett Packard Enterprise and Nimble!
Materials:
Source: habr.com
