AppNeta | The Path

Posts Tagged ‘Monitor network performance

The link between Apple computer and the PathView microAppliance™ might not be Apparent, but it is there.  

The microAppliance is the latest innovation from Apparent Networks, and it is turning out to be a game changing device.  Apparent Network’s flagship AppCritical and PathView network management tools, has long been a technology leader in the area of Network Performance Management.  During the past 10 years, some of the world’s largest companies and IT outsourcing firms have counted on us to deliver precise network performance information about far-flung wide area network that is critical to their business. We are excited to announce that we recently repackaged our core technology into the PathView product line, which is available as software as a service (SaaS) and downloadable software. The PathView solution was designed for individual users rather than enterprise wide deployments.

The game changer of late has been the microAppliance.  Why? Simple…Performance is a perspective-critical task.  Customers generally want to know the performance of a network “from point A to point B.”  This has traditionally been “from my service delivery data center to my remote office.”  AppCritical and PathView are best in class at actually understanding ongoing network performance in a scenario like that.  They use the same core technology to (pre)assess, continually monitor and troubleshoot network issues.  They both use a software component known as a “sequencer” (also embedded within every microAppliance) which uses micro-packet trains to measure performance without impacting ongoing application delivery.  Very cool stuff.  

But when the question is “What is the performance from my Detroit sales office to my LA sales office?”, things get a little more interesting.  First, why would someone care? They would care because they might now be using more updated and efficient network technologies such as leased MPLS, and rather than the old “hub and spoke” networks, MPLS allows site-to-site traffic. This is very important for an applications such as VoIP for instance.  Customers also want to know detailed performance information about network devices from the perspective of the remote site, so having a performance management element onsite make perfect sense, except for 1) cost 2) deployment hassles and 3) ongoing management.  

This is where the PathView microAppliance comes in.  

It is a zero remote administration device (just plug it into power and the network, and you’re done), and it is free with a subscription to PathView Cloud (pricing starting well under $100 / month). So, game changing?  You bet.  Now you can easily have detailed performance information at….your remote retails site…your remote health care facilities…your off site development offices…your sales offices….(you get the picture?).

So where does Apple come in?  

The microAppliance is really a work of art in of itself.  It is a near 100% solid-state device (no moving parts) that leverages the latest in “System On a Chip” (SOC) architectures. It consumes about 5 watts of total power, requires no special cooling, measures less than 4 x 3 x 2 inches and weighs in at about 14 oz. And, thanks to the highly tuned PathView software that runs on the powerful CPU, ample flash and RAM capacity, we can accurately measure up to 1 Gbps networks at full wire speeds!  All of these things are incredibly affordable (wait for it…) thanks to the HUGE success that Apple has had with iPods, iPhones, iTouch’s, iPads, etc, etc. Millions of devices have driven a market for new price points on critical technology components.  Good cheap components, in a great performance management solution, at price points you aren’t going to believe. Try it for yourself – for free.

Thanks Apple….

Advertisements

Earlier this week, our Cloud Performance Center (CPC) detected a connectivity loss that occurred at Amazon’s EC2 data center in northern Virginia. The CPC is powered by our PathView Cloud tool, which monitors network path performance into and out of the key datacenters of several major cloud service providers across the U.S. As part of the free CPC service, we issued a Performance Advisory later that same day, informing cloud service consumers and other interested parties about the location, nature and duration of the problem. 

The point of this whole exercise is not some kind of ‘gotcha’ network management. In fact, Amazon and the other providers we monitor do a great job of providing reliable services. The point is that when the Internet is involved, stuff happens.

If your company is one of the many that has already moved critical applications into the Cloud, you need to know about that “stuff’ quickly and whether or not it’s affecting your operations.  You need to know what is going on not only at your provider’s datacenter, but more importantly, between your Cloud provider and your customers and employees. 

To gain that critical insight, you need the network-path perspective.  That’s what our PathView Cloud tool and our CPC resources deliver. 

Cloud computing is a great technological advancement that has already delivered lots of value.  But we’re only scratching the surface of what the cloud approach can do for us.  For cloud computing to live up to all the hype, we all need a greater understanding and awareness of the network paths that go to, through and from the cloud. 

Apparent Networks is already there!

In my last blog, I talked about some interesting parallels between virtualization and IP networking – both are essential technologies and yet suffer from poor visibility.  As virtualization has matured, the industry has scrambled to compensate for its blind spots.  However, most of the proposed solutions haven’t been particularly creative or effective.

Continuing to think about the parallels, an interesting possibility recently occurred to me:  Can PathView’s unique approach to probing and monitoring IP networks make a difference to virtualization?

PathView’s point and shoot magic applies equally well to network paths that terminate in a virtual machine at one or both ends.  The nature of IP packets means that, by our definition, the network path now extends up into the physical host, through the virtual network of the hypervisor (that includes the virtual switch), and well into the virtual machine.

This is the same path that any application traffic must take.  Any test packets traveling along this path will meet the same degradation in performance as those of the application.  By analyzing the behavior of test packets, it is possible to isolate the effects at Layers 1, 2 or 3 that impact end-to-end; all without having any special access to any of the intermediate devices, or even instrumenting the target.

How does this help monitoring virtualized systems?

If the end-to-end network shows clean, fast network performance, then there is no need to instrument or interrogate anything in-between or on the virtual host.  If there is a problem, it is possible to isolate the location in the Layer 3 network, determine if it is at the VM and even distinguish what sort of problem it is.  At this point, you know that you have a problem, whether it’s the network, VM or application, and even what kind of issue.

But that’s when it hit me – the virtual machine is handling packets as they pass through the virtual switch, the virtual NIC and driver.  And just like with any other network device, the overall health of the virtual machine is effectively impressed on the packets themselves.  As a virtual machine runs out of resources, or otherwise has difficulty performing, it will also have difficulties handling packets.  Packets will experience increasing levels of jitter and decreased capacity on the virtual segment of the end-to-end path.

In other words, the network path acts as the canary in the virtualized coalmine, signaling when the virtual environment degrades.

In applying the PathView’s end-to-end network path analysis to virtualization, several things are true:

  • only the end-point IP address is needed
  • end-to-end network performance means good VM performance
  • if the virtual machine begins to degrade, it will show up on the packet timings
  • if the virtual application moves, so does the monitoring path
  • network path monitoring scales much better than device monitoring
  • feedback is available in minutes, not hours

Not only does this approach simplify how you monitor your virtual infrastructure, and directly assures application performance, it solves the problem of monitoring 3rd party cloud infrastructure too.  Instead of having to put blind trust in your cloud provider to do the right thing, you can continuously monitor the performance remotely.  And call them on it if you need to.  It seems almost too easy…

In my next blog, I’ll write about what network monitoring can (and can’t) tell you about the state of virtualized applications.


Follow us on Twitter!

Error: Twitter did not respond. Please wait a few minutes and refresh this page.

%d bloggers like this: