I had the pleasure of waking up around 7:30 yesterday morning to have coffee with Juan Cabezas, VP Marketing and Sales of nVision Software, yesterday morning. You read correctly. 7:30 in the AM. That’s how much I like systems management.
When it comes to The Hookup, as ever, my old BMC pal Israel Gat is the man: he email-introduced Juan and I. Israel is one of those networking spanning layers that seems to know or have worked with just about everyone in systems management that I met: Jeffrey Snover for example.
Juan and I had a couple hours of good and fun conversation about marketing software, small engines, GE, systems management, Houston, and the trouble with us Southern boys (he of Florida, me of Texas) fitting in in the Northwest (mowing roofs in Seattle).
nVision’s AppVision 5.0
We talked, of course, a great deal about nVision’s AppVision 5.0. To put it in short hand, AppVision is a way to instrument Java and .Net applications to start gathering information internally on those applications. Instead of only monitoring from outside of the application, AppVision goes in (with AOP-ish byte-code I’m guessing) and starts tracking everything that happens inside the application. There are non-instrument ways to monitor as well for non-Java and .Net aspects of you system.
The story from there is best practices in systems management: first create a baseline so that you know what “normal” is, monitor for exceptions to that baseline, take a snap-shot of what’s happening at the moment, and analyze that snapshot both programatically and with human eyes to figure out what’s going wrong. There are some management abilities as well to attempt an automated recovery.
There is, of course, a console, workflow/rules built around event detection, and notifications.
Identify and Wily pop to mind as similar applications, of course.
Automated Debugging, Marketing
As we talked more, I realized that AppVision could potentially be the “automated debugging” (the phrase “headless debugging” might be better, if more technical) that I always wanted when I was in my support rounds at BMC. When an application is crapping out in the field, and all the usual tricks and log hunting come up with nothing, what you really want to do (but can’t) is attach a debugger to the application and start rooting around for things that look whacky. You have a baseline in your head, and you want to find the exceptions.
We talked a great deal about community based ways of marketing AppVision, and narrowing down on this automated debugging was one of my key suggestions. AppVision could take a slice of it’s functionality — say memory leak detection and/or socket monitoring — and create a free tool around it. Support people could use this tool when they’d like to debug in the field, sending the AppVisionDebugger (to pick a name) to their customers to use a probing tool.
Indeed, many of the customer stories we went over did just that for in-house applications. One customer had purchased a large piece of Java middle-ware from a vendor, adding in customized code to run on-top of it. When failures came up, the large vendor would deflect the finger-pointing by saying, “it must be in your code; prove otherwise, and we’ll open a ticket” (I’ve been on that side of the fence countless times ;>). So, the customer used AppVision to look around inside the application, found that it was the vendors code, and did their own finger-pointing deflection back up to the vendor.
All you 3rd level support people out there can see how any help deflecting the finger-pointing would be helpful.
Of course, once people started using the free version, some to many of them would see the value in that tool, and start spending money. It’s the freemium model we’re always talking about here at RedMonk.
Another piece of marketing advice I gave was to take a page from Atlassian and Clover‘s book and give away free copies to prominent open source projects. Open source projects do plenty of “3rd level support,” and tools that have low-barries to entry of cost and deployment/run time are attractive (those are attractive anyone, really).
As you can imagine, it’s the same freemium model, but with viral effects as well. If OSS projects start sending out AppVision agents to help them support their products, all those people on the recieving end get exposed to AppVision. Some to many of them will, the hope is, think, “huh, that looks like it’d be useful for us. Time to get the manager to shell out some doe.”
In their general form, of course, these two strategies are applicable to all software marketing, esp. if the software is a tool or utility.
More to Come
Now, full disclaimer, my understanding of AppVision is from what I’ve read and talked about with Juan, I can’t vouch for the product technically, just for the the story. That said, I have a more technical overview lined up, and I asked Juan to send me a copy so that I could poke at it.
It does look interesting however, so I’m looking forward to checking out more.
Disclaimer: BMC is a client.
Technorati Tags: nvision, appvision, freemium, austin, redmonkadvice
Just curious – were you able to evaluate AppVision? What are your impressions, and is it easy to implement?
Running a Support organization for a "middleware company", I can see the usefulness of a product that can help identify where the problem resides quickly.
Let me know what you think…
Regards,
David Mazursky
I haven't the chance to do a "labs" type of testing, but I have played around with it. It seems like a good way to diagnose problems, but the usablity could be improved a bit. I think that's more of a 1.0 type of problem more than anything else. Using Eclipse would be a nice step.
I'd be curious what your thoughts on it are. Feel free to post here or email me ([email protected]).
Well, what I’ve seen looks interesting, but we haven’t looked at it in detail. I’m looking at it from 2 perspectives: 1) using it for support for our partners/customers, and 2) recommending or integrating it as part of an overall solution for availability we provide.
Our customers use our product, everRun, to protect critical applications from Windows server problems, a tool like this might be a good complement to diagnose application problems when they do occur (manageability and reliability being factors contributing to overall availability to a user).
Since we’re in exploration mode right now, are there similar technologies to AppVision you’ve encountered?