On the excellent DevOps mailing list this week I stumpled across two interesting entries on monitoring tools. Note that I am not writing this to attack anyone, but to share some observations that I have relevant to our own tools and where we can help. I realize that our own platform vCenter Operations Manager (and packaging variants such as vSOM and vCloud Suite) caters sometimes for another type of customer than the typical devops crowd. As an early linux fan and ex-Sunnie I still care deeply about open-source (oops did I really say that ?) and projects like Nagios or Zabbix have their place. Maybe…

The first is a survey done by Dataloop.IO, a company that is developing a new SaaS monitoring tool. In a blog post they comment on a survey they did of 60 customers in the UK and their monitoring habits. Some of their findings captured my interest:

• People use a pletora of tools and build-your-own: Yes, that is what we see too. People use tools because they are “free” (if your time is free) like Nagios, because they got them from a vendor as a – limited free – tool to use (CIM, MS SCOM,…) or bought them for a limited purpose (Veeam,…). For anyone using vSphere as a platform I think vCOps is a strategic choice. You will see that it will develop more and more into a platform and integrate with other products. Think cloud provisioning asking vCOps on which cluster there are enough resources to deploy. And whether you have vCenter and 100 VMs or want to monitor a vCloud service at the vapp level, this is the one tool that you use. vCOps is also becoming more of a platform with Hyperiq merging in: you can now monitor hundreds of services that are not VMware specific, and best of all vCOps discovers the relationships and shows you for example on which VM Windows is running with your MS SQL database. Magic. And did I mention more and more management packs are being developped to bring other data in vCOps ? Brocade, HP 3PAR, Dell storage, EMC, NetApp, Oracle,… Have a look on our management marketplace.
• micro-services: Again I think people that are now building different tools for their departments are well served with vCOps. You can monitor centrally, but everyone can manage their own view using custom dashboards you build in minutes. We have customers monitoring java apps centrally and offer internal customers their own dashboard of their app without interfering with others or seeing their data.
• alert storms: Well, we have been telling you about this one for a long time…. You need a modern tool like vCOps that does dynamic analysis of your data and sends you intelligent alerts when needed. We learn from your data and do trending…

The second entry corroborates a further finding from Dataloop.IO: that most people still use Nagios. Andy Sykes from Forward3D argues in a Slideshare presentation that we should stop using Nagios for a number of reasons  – overall  because it is outdated. But he proposes to build another system that is composed of a number of tools… See the Dataloop.IO study….

Again I think the case is for vCOps. At a large European institution we have not so much replaced Nagios as we have enhanced it. Where their old system was complex and hard to change, we now import critical data from Nagios (around ntp…) and display it in custom dashboards in minutes. The same tool is used with other dashboards to monitor vapps in the vSphere environment and on top of that vCOps does it’s analysis magic on imported data too, so they have an idea of what the normal trending of ntp offset is, something you cannot do with Nagios.

vCOps will not replace all other tools, but for customers adopting a software-defined datacenter model it is a strategic tool that can serve the infrastructure department and it’s customers, serving everyone relevant data and alerts from one tool in different forms.