Monday, December 17, 2012

Taking our story down-under

The winding waterfront of Melbourne
A week ago, James Dollar, the founder and CTO of Uplogix had a chance to address the CIOs of one of the largest corporations in the world at a gathering in Australia. It was a chance to introduce a global audience to the concept of local management, as well as show them why it's something they need to know about. Here is a summary of what he had to say.

Today, Connectivity is King
The trend is toward smarter and more connected devices and systems. Examples range from the obvious, like the Bring Your Own Device (BYOD) movement, to the less conspicuous like unmanned energy transmission sites in the middle of nowhere connected by satellite links.

Under our noses, but not necessarily on the radar for networked systems are the smart devices like an MRI machine that self-monitors and "calls-in" for service as needed. Even the ubiquitous ATM where you pull out cash is likely a system made up of redundant wireless links and individual components for check scanning, cash dispensing and security monitoring -- all reporting back to a NOC somewhere on a network connection.


With the reliance on connectivity growing daily, business requirements for that connection are becoming more stringent. The key needs are:
  • Uptime – These devices are only useful if they are working, and connected with an acceptable QoS
  • Security – Absolutely critical to operations, customers
  • Acceptable Operating Cost – Increasing numbers of complex and critical devices is a formula for support costs to skyrocket
This last bullet is the kicker -- generally, if something is important-enough, and money is no object, layers of redundancy and lots of people can almost always conquer reliability issues. But rarely is money no object, and we're talking about a need for reliable networks almost everywhere and with billions of network devices. (Cisco predicts 50 billion by 2020!)

You can't put highly-trained people everywhere, all of the time, so operations are centralized. This introduces new problems:
  • Complexity and triaging of issues
    Everything goes into one big bucket for observation; collecting and processing vast amounts of monitoring data
  • Downtime and truck rolls
    Relying on the network itself to manage networked devices. It's like driving on a spare tire -- if it goes out, you are walking.
  • Security and compliance risk
    Pushing updates is difficult and sometimes risky, so the norm is to wait on updates until they are required, standardizing on major revisions
So what's the answer?
Distribute and automate the device management from a local perspective. Put intelligence out with these devices to manage them autonomously, and provide support tools for problems that are too difficult for automation.

Today, most "management" tools are really monitoring and filtering tools. The difference between them and local management is the difference between knowing there is an issue and FIXING the issue.

What do you need to have a local management solution? There are several key components for a successful platform including:
  • Directly connecting to devices
    See everything an on-site technician would see, more reliably and without impacting the network
  • On-board processing & storage
    Monitor, evaluate and take actions based on the run-book, back up configurations and OS files onsite
  • Secure out-of-band access
    Provide an off-network connection for more complicated problem resolution by a remote technician
  • Role-based access and audited management
    Eliminate potential for abuse and ensure policy compliance
  • Complement existing centralized management tools
    Feed the NOC with information, backfill for audit and compliance
The presentation went on to show some example of how this idea of local management answers many of the challenges of the increasingly connected world of today, as well as a pathway to an even more advanced world tomorrow. Check out the slides above!