Monitoring from the Cloud – APMaaS
Recently we have been getting involved in Service Engage from IBM. An offering that allows customers to run Enterprise Systems Management functions as a cloud service. The beta currently offers the following areas to be run using a SaaS model.
- Performance Management
- IT Service Management
- Workload Automation
- Internet of Things
- Mobility Endpoint Management
As an IBM business partner of many years, the idea of offering some or all of the above aaS has been of great interest to us. Currently we provide some of these on premise, on the cloud, aaS and some we have found difficult. This blog aims to discuss APMaaS ( Application Performance Management as a Service).
Moving a service that is traditionally on premise to the cloud is always met with trepidation and concern. Customers view putting their monitoring infrastructure on the cloud as new, and this will result in some issues to overcome – usually around security of data and levels of service. However, I believe that this is the way forward for many customers purely because of cost and ease of use. I have seen enterprise systems management environments from very small to extremely large, covering vast systems with huge running costs. They usually always end up becoming extremely complex and grow out of control.
If I was in charge of a large enterprise systems management infrastructure, my first aim would always be providing a reliable service that delivers, very closely followed by how can I run it smarter and cheaper.
I don’t actually want all the hassle of building and maintaining infrastructure or exorbitant staff costs. I also don’t want or need staff redeveloping fancy front ends that cost a lot to maintain. These can and should be left to the vendor, partner or a very small team of specialists.
What I need is the ability to run APM quickly, simply and pay for all that infrastructure and overhead costs as a service. Potentially saving millions. Here I attempt to show how quickly and easily I can start from nothing to get a monitored service in a very short space of time.
Starting from Scratch
With no infrastructure and no desire to acquire skills I want to monitor some systems. The first step is to get registered to the beta, once this is done you can login to Service Engage at https://serviceengage.ng.bluemix.net/
This will bring you to the Service Engage page allowing you to request a free trial of all the above SaaS offerings. There is also a lot of information including videos, blogs, documentation and so on. Our focus in this article is on APMaaS.
Having requested a trial, you will wait about 3-4 hours while your environment is provisioned on SoftLayer, IBM’s global cloud infrastructure. You are then notified of a Performance Management service being available to you. Selecting the My Services link will give a list of all cloud services you have requested.
Here I have 1 service, Performance Management. Selecting the Agents link from here gives me the ability to download agents and view installation instructions.
The installation of these agents are simple shell scripts or batch files that are run on your monitored hosts. These are so straightforward there is no need to document the steps. These require very little configuration and no knowledge of any IBM Tivoli Monitoring or Smart Cloud infrastructure. All communication is based on SSL and signed certificates from your subscription (more details to follow in a subsequent post).
Once your agents are installed, simply Launch into your APMaaS environment using the Launch button.
And you are presented with your APMaaS Getting Started homepage.
Now we are monitoring systems in the cloud using a proper SaaS model. Very easy to deploy, very few skills required and very little infrastructure cost. From here we can simply look at our performance dashboard, and start to build applications to monitor.
Building an Application
Now that I have agents configured and running I can very quickly build an application to group my systems into a service view. First of all create the application by giving it a name and if required some transaction data from say response time agents that are listening to requests from a website.
Next choose resources you wish to be part of this application. We are going to choose some operating systems, a MySQL instance and a WebSphere Application Server instance.
This gives us a very simple application view containing Transactions, Components and subsequent Events relative to my application. I can now drill down into these different parts that constitute my application to identify problems or look at specific resources.
My application view has a status overview allowing me to very easily see Transactions and Components. I can then drill into the components by type or all together.
Viewing Monitored Resources
The End User Transactions are HTTP/HTTPs transactions that I am monitoring from the response time agent. The configuration of this agent is very simple with just a NIC address and port to be defined.
The components that are grouped to form my application can be viewed all together or by component type. The following screen shot shows Operating System, Database and Application Server metrics on the one page.
Alternatively I can look at only the Operating System metrics (Windows in this case) that constitute my application.
I can drill down to get more detailed information about a single instance. Here I see CPU, Disk, Memory and Network information in a very easy to read display.
Similarly for a component such as WebSphere Application Sever (or Ruby, MongoDB, MySQL, Linux, WebSphere Portal Server etc).
There are many levels of depth that can be attained by selecting the component you wish more information about.
Currently all events are taken from the situation event console within IBM Tivoli Monitoring, however future releases may include OMNIbus event lists.
Having spent many years building, maintaining and developing monitoring infrastructures and seeing how expensive they become to maintain, the difficulties organisations have in retaining skills, the overall lack of simplification and the inability to make sense of complexity, the first signs of being capable of doing all this as a SaaS offering I see as a game changer. Here are some of my initial thoughts.
- “I was able to download and deploy agents to monitor Windows, Linux, WebSphere, MySQL and Transaction Response Times within 30 minutes.”
- “A few clicks to build an application to monitor my services. Very tidy and modern interface, easy to use and understand.”
- “Monitoring aaS has always been hindered by infrastructure communications. This APMaaS seems to have overcome this. A very neat and exciting development.”
- “I have managed to monitor some systems without needing to install TEMS, TEPS, TDW, TIP, Java WebStart, Databases (and more!). I don’t even need to understand them as it is all done as a service.”
The ability to save any organisation on a) staff costs and b) infrastructure has got to be high on anyone’s agenda. We have found it relatively easy in the past to help customers reduce a) but reducing b) has always been difficult because it means moving infrastructure outside and brings fears surrounding security. Using SSL and signed certificates seems to overcome this.
As this is a beta, there are rough edges, however things look very promising. As cloud starts to deliver more over the foreseeable future, organisations will look to adopt this type of model.
The smart organisations will be those who are already looking at this type of delivery for traditional monitoring and event management solutions.