Eating our own dog food, or sipping our own champagne

Eating your own dog food. The phrase comes from the early television advertising genre when people would ask, but will the dog eat the food? Today it’s categorized as a computer jargon phrase, well-documented in a Wikipedia entry, describing how software companies and other industries try out their own products, putting themselves in their customers’ shoes. I assure you that at BMC, our IT group often pops open cans of our dog food, or sips our own champagne, as Thomas Siebel prefers to call it.

My favorite essay on the topic has to be Joel Spolsky’s “What is the work of dogs in this country?” essay from 2001. Read both the Wikipedia link and the essay for all the nuances and pros and cons of eating one’s own dog food. I especially like Joel’s example of how the Juno executive wanted six pop-up ads until he experienced it himself and then backed off to two pop-up ads.

Similarly, I have heard people ask over and over again, this BSM stuff sounds great, but are you really doing it internally, BMC? That type of question is the heart of eating one’s own dog food. “Sounds great, but have you put it into practice? Show me.”

In the spirit of documenting how we eat our own dog food, I recently completed a white paper, “Implementing Resource Management Using Business Service Management Principles.” It’s about how our research and development lab schedules server resources for testing. This scheduling is no small task, especially in an Agile development environment with iterations that might go for 2 weeks or 4 weeks. The products we test also may need to be tested on 14 different platforms. I think the team used a half dozen BMC products and intend to use even more in the future, such as the advanced discovery and provisioning tools we have available. Products consumed so far, with more on the way:

• BMC Performance Manager for Databases
• BMC Performance Manager for Servers
• BMC Performance Manager Portal
• BMC Remedy Action Request System
• BMC Remedy Asset Management
• BMC Remedy Change Management

So take a look at the paper – no registration is required – and let us know what you think. This IT group, including talk.bmc’s own Steve Carl, is constantly looking forward to make their processes even more business service centered.

Subscribe to the comments for this post

Follow up for ITIL and monitoring

Shortly after posting my ITIL and monitoring scenario about BMC Performance Manager notifying CONTROL-M that an SAP job was failing, I got a great response about how CONTROL-M even takes it a step further towards ITIL ideals with the Batch Impact Manager module.

Just read your blog. I wanted to update you on a couple of items which help to fill out the drive to ITIL standards achievement in the operations and monitoring environments.

BMC CONTROL-M has an additional module called BMC Batch Impact Manager. This module allows a user to reduce a large flow of jobs to a single service instance and monitor those services critical to the business. Is there any other kind? When a service is predicted by Batch Impact Manager to fail its due out time, Batch Impact Manager issues alerts to operations and provides an interface to Service Impact Manager to focus attention on bringing the service back to normal. Therefore if an SAP process was included in that critical service, as soon as the non-availability of the SAP process impacted the service, CONTROL-M and or Batch Impact Manager would start to squeal.

Great stuff! I’ve just added some links and spelled out the acronyms, otherwise unretouched. The embarrassing thing is, I knew about this module but hadn’t made the connection. Thanks Ronnie for bringing it to my attention!

Subscribe to the comments for this post

ITIL and monitoring

I’m relatively new to BMC Performance Manager but I’ve learned a lot in the past year or so about the capabilities and how it fits into the Infrastructure and Application Management Route to Value. So this post talks about a white paper my co-worker Bill wrote, BMC Performance Manager – Cornerstone of your ITIL Implementation. (Look at that, our white papers no longer require a webform registration before you can download them!)

I like that he includes realistic scenarios. Scenarios help me understand the real issues and problems that need to be solved by businesses. With BMC Performance Manager for Servers, you can manage the health and performance of different operating-system environments and applications. BMC Performance Manager helps you close in on ITIL Availability Management objectives and, when used with other BMC Software products, also helps companies with other ITIL processes, such as Incident Management and Problem Management. This white paper describes how you can use BMC Performance Manager for availability management. Here’s an example scenario from the white paper.

A company uses SAP for supply chain management and BMC CONTROL-M for scheduling SAP jobs. When there is an unscheduled outage of the SAP system due to excessive paging that makes it unavailable to suppliers, BMC Performance Manager automatically detects the problem and initiates recovery procedures.

Unless BMC Performance Manager reports this unavailability, CONTROL-M is not aware of this outage and would continue to schedule jobs to SAP.

Because the SAP system is not active, business processes managed by CONTROL-M could fail, resulting in long recovery time that impacts SAP users’ productivity. When an unplanned failure of an SAP application is detected by BMC Performance Manager and then reported automatically to CONTROL-M, the operator can use the data provided by BMC Performance Manager for Problem Investigation.

What do you think about this scenario? Sound like something that could happen in your environment? Let us know if we’re on the mark, and share any harrowing experiences you have.

Subscribe to the comments for this post

How to create help files for custom BMC Performance Managers

All this is documented in the Controlled Availability release of the BMC Performance Manager Software Development Kit (SDK), but I thought I’d write it up here as well, hoping it’s helpful. Contact your sales rep if you want more information about the SDK for BMC Performance Manager.

BMC Performance Manager is a product with the ability to be extended, allowing you to write your own custom monitoring tools, called Performance Managers. If you write a custom Performance Manager, you’re going to want a help system to go with it, so that your users know which parameters are monitored with your tool. And if you want your custom Performance Manager to be certified by BMC Software, a help system is required. Here’s an overview of writing that custom help system including sample files.

1. Write an HTML file for each application class and the parameters within that class, nested like this set of sample HTML files for the BMC Performance Manager for Citrix Presentation Server . For example, the farm application class contains parameters like “logged in users” and “disconnected sessions,” with each parameter documented in separate HTML files. These files contain Dreamweaver template code but should be useable with any HTML editor.
2. Place your Help content files in your Performance Manager Maven project in the following location: …/META-INF/help/browser_help/.
3. Open your application definition XML file and add these Help elements in the application definition: help-group-definition or help-group-reference, and help-item. You need unique attributes msgkey and name on both these elements. Here is an example code snippet:

4.  <application-definition name="patsdk-ri-service">
   
   <display-name>Reference Implementation Application</display-name>
   
   <description> Instances of this application-definition can be used
   
   to monitor the availability of many common network services
   
   such as HTTP, Telnet, FTP, etc.</description>
   
   <help-group-definition msgkey="riapp.intro.displayName" name="intro">
   
   <display-name>Patsdk RI Application</display-name>
   
   <help-item msgkey="riapp.intro.about.displayName" name="about">
   
   <display-name>About</display-name>
   
   <file>riapp/about.htm</file>
   
   </help-item>
   
   </help-group-definition> <help-group-definition
   
   msgkey="DRCIT-Farm-Container.desc.book"
   
   name="DRCIT-Farm-Containerbook">
   </help-group-definition>

5. Add Help-related fields to your project.properties file (located in your Maven project folder). There are a couple of commented lines that you can uncomment by deleting the # sign at the beginning of the line. Set solution.product.code equal to your PAR file name (without the .par extension), and use a sample category such as database, networking, etc.

6. solution.product.code=PRD
   
   solution.help.category=networking

7. When you build and deploy your Performance Manager, the .htm files that you created are compiled into cross-platform browser-based help. These files and the Help content are packaged in a .jar file which is then put into the .par file and deployed to the BMC Portal Help repository server. Automagically.

Sounds pretty straightforward, and we’ve done this internally for a few Performance Managers already, which is why I wanted to share some sample files with you. Let us know how it works for you.

Subscribe to the comments for this post

Getting BMC parameter information that you can manipulate

There’s a tool called the Parameter Reference Database on our support site that lets you export the PATROL Knowledge Module parameters to a CSV file so you can re-mix them as you want.

This Parameter Reference Database tool for exporting the KM parameters looks really cool to me so I wanted to be sure you all know about it too. One neat thing about this type of documentation is that updates are not tied to product release cycles. It’s much closer to the single sourcing holy grail where you change your source file and that change can be propagated through the whole system with a few button clicks. If you’re more of a driller-down type, go to the Support Central page, and then click Product Lists & Manuals, and then click Parameter Reference Database.

Once you’re on the Parameter Reference Database page, select the products whose parameter information you want to download. Here’s a list of the products.

Next pick the version number, click another button or two, and you have a CSV file! Here are the categories for each parameter that the tool gives you in the CSV file.

Let me know how you and your teams are re-mixing the content. I’m envisioning database queries, XML output, other types of outputs that can be created with a CSV file. What’cha got? What other reference material might be good to output in this way? I’ve done some work with XML output for error messages so that’s another type of content I’m interested in single-sourcing. Bring it on.

Subscribe to the comments for this post

Photos from the forum

I took some photos at the BMC Forum in Dallas October 2005

I had some fun with my digital SLR pretending to be a photojournalist. Here are some photos of activity at the forum.

Chip and Stephen work for Dell and Temple-Inland, respectively, in Austin, Texas, and they’re brothers. I had to ask them if I could play paparrazi and snap some shots.

Checking email between sessions like a lot of us were doing.

A lively discussion in the hallway requiring hand guestures and everything.

Another discussion after the Marimba 101 lesson.

Here’s a partial shot of the expo area where you can go see products in action.

Subscribe to the comments for this post

Tuesday BMC Performance Manager session at the BMC Forum 05 in Dallas

Reporting from a conference room set at 60 degrees Fahrenheit, here’s your roving blogger reporting from Dallas

Blogging live is harder than it sounds. Fortunately the wireless connection is behaving in two of the session rooms I’ve been in so far. But, beyond the technology (which is the easy part), it’s difficult to take notes and figure out what to report on. So here goes. Let me know if you’d like to hear more.

There are plenty of sessions to choose from and at least five tracks. This morning I went to the BMC Performance Manager Roadmap and Strategy session with about 35 attendees. Sean Duclaux started with a trick question by asking for a show of hands. How many PATROL Express customers? (a few) How many PATROL Classic customers? (a bunch) How many BMC Performance Manager customers? All! We’ve changed the PATROL product name to BMC Performance Manager. Of course with a product evolving like this, lots of questions ensue. I’ll try to capture the questions and answers here.

Q: How do you decide which to use, agent-based or agentless monitoring?
A: Based on collection policies that you set, the agent might deploy automatically, perhaps by pushing a lightweight local presence onto the computer to be monitored. More on this below the question/answer set.

Q: What kind of pricing is available for people who are already invested in the PATROL Classic product line?
A: The licensing scheme has been completely redesigned in a few ways. One is that there’s a CPU metric, so if you want to monitor a server, it doesn’t matter if it’s Windows or UNIX or Linux — you can switch between them. Also there are tiers of deployment that are simplified, such as a departmental license. I’m sure I’m missing some layers here but the overall answer is that PATROL Classic is not going away, but you will see infrastructure cost savings as you upgrade and decommission old infrastructure.

Q: What technological help is available for upgrading our KMs?
A: The BMC Performance Manager SDK was just released in August and you can request it (it comes free with BMC Performance Manager). With this SDK you can create application classes and XML config files that will pick up all the info that your KMs do (as long as it makes sense to do so), and there are third party implementers being trained on the SDK right now. (OTL is in Austin this week for training, apparently).

Q: What about about the install footprint — how much disk space for this lightweight local presense?
A: It shouldn’t be a big space hog. Just looking at my own Marimba client install I’m seeing a less than 50 MB install, and Marimba is the one that gathers the most information, not a lightweight local presense. I’m guessing lightweight should be MUCH less than this.

Q: What about bandwidth, will it fill up my network sending data back and forth?
A: This is all configurable, but typically only when an event is raised will it be sent back. Of course if you’re going from PATROL 3, which apparently didn’t send data anywhere (I’m no expert on this but that’s what was said), you’re going to see a difference in network capacity.

Q: The Million Dollar Question (according to a guy in a UNIX-only shop) — will the RSM (Remote Service Monitor) run on a UNIX box or is it Windows only?
A: The product manager and architect are arm wrestling over that right now. The basic answer is that we (well, the architect) wants to do everything, but … a Windows RSM can monitor both Windows and UNIX, but a Solaris/UNIX RSM can only monitor UNIX, so we need to know whether that’s worth building — does it fit into the environment that you envision? UNIX doesn’t exactly listen well (ok, at all) to perfmon information, for example, so there’s no monitoring of Windows with a UNIX RSM.

Q: Will a lightweight local presence (LLP) incorporate auto recovery actions?
A: Even if you are managing a solution remotely, as long as dynamic connection can happen, we’ll let you do recovery actions for remote connections (not til after December though.) PATROL Express can do remote restarts right now.

The gee-whiz factor for me with the new direction is the combination of agent-based and agentless options. Both are available now with a single view point, meaning your PATROL Express data can be viewed alongside your PATROL data. You can apply a policy to determine whether you monitor something with an agent or not. Here’s an example of a policy application out of the BMC white paper, “Effortless System Management.”

Policy example: If a small file/print server is reassigned to serve the office of the company’s chief executive,
the IT staff may decide that it wants an autorecovery capability on that server. The IT staff simply sets the
new performance management policy for that server, and BMC Performance Manager makes the
appropriate changes, which may include pushing an LLP out to the server.

Another recurring topic so far is compliance efforts such as Sarbanes-Oxley, HIPPA, Basel II. As Sean put it, “You don’t want to see your CEO on the cover of a magazine in an orange jumpsuit.” So, if Sarbanes-Oxley or other compliance efforts are your concern, figure out how to get your policies in place. I’m hearing this over and over.

All the presentations are available with a username and password, so if you’re attending, here’s the site to download the presentations. Your packet has the username and password.

Subscribe to the comments for this post