Huh? What’s the Network Have to Do with It?

By Nate Schnable, Sr. Solutions Architect

Having been in this field for 17 years it still amazes me that people always tend to forget about the network.  Everything a user accesses on their device that isn’t installed or stored locally, depends on the network more than any other element of the environment.   It’s responsible for the quick and reliable transport of data. That means the user experience while working with remote files and applications, almost completely depends on the network.

However, this isn’t always obvious to everyone.  Therefore, they will rarely ask for network related services as they aren’t aware the network is the cause of their problems.  Whether it is a storage, compute, virtualization or IP Telephony initiative – all of these types of projects rely heavily on the network to function properly.  In fact, the network is the only element of a customer’s environment that touches every other component. Its stability can make or break the success and all important user experience.

In a VoIP initiative we have to consider, amongst many things, that proper QoS policies be setup –  so let’s hope you are not running on some dumb hubs.  Power over Ethernet (PoE) for the phones should be available unless you want to use bricks of some type of mid-span device (yuck).  I used to work for a Fortune 50 Insurance Company and one day an employee decided to plug both of the ports on their phone into the network because it would make the experience even better – not so much.  They brought down that whole environment.  Made some changes after that to avoid that happening again!

In a Disaster Recovery project we have to take a look at distances and subsequent latencies between locations.  What is the bandwidth and how much data do you need to back up?   Do we have Layer 2 handoffs between sites or is it more of a traditional L3 site to site connection?

If we are implementing a new iSCSI SAN do we need ten gig or one gig?  Do your switches support Jumbo Frames and flow control?  Hope that your iSCSI switches are truly stackable because spanning-tree could cause some of those paths to be redundant, but not active.

I was reading the other day that the sales of smart phones and tablets would reach approximately 1.2 billiion in 2013.  Some of these will most certainly end up on your wireless networks.  How to manage that is definitely a topic for another day.

In the end it just makes sense that you really need to consider the network implications before jumping into almost any type of IT initiative.  Just because those green lights are flickering doesn’t mean it’s all good.

 

To learn more about how GreenPages Networking Practice can help your organization, fill out this form and someone will be in touch with you shortly.

  • http://www.heroix.com/ Susan Bilder

    The network is the foundation for everything which is why a network monitoring solution should be in place. You don’t want to take a reactive approach to monitoring the network. It’s important to be alerted that there is an issue and fix it before end users are even affected.

    • Nate Schnable

      Hi Susan

      Thanks for your post – and I definitely agree with you.

      Having some SNMP monitoring (and flow-based too if possible) definitely allows us to be proactive rather than reactive. This can either be in-house or cloud-based.

      We use it internally and also offer it as a service for our customers.

      If you want to read about it please feel free.

      http://www.greenpages.com/business-services/greenpages-cloud-management-as-a-service-cmaas

      I would love to know what you are using – thank you for reaching out!

      Nate

  • http://twitter.com/souravoncloud Sourav Ray

    Agreed with Nate. But Nate according to you, as a Solution Architect what parameters do we need to consider to provide effective communication over internet, either cable or wireless??

    • Nate Schnable

      Good question, As we know there is no real QoS on the Internet we would have to tackle this differently. If you had multiple data centers you could implement some global DNS load-balancers. This would funnel users traffic to the closet/best performing data center. Also would provide redundancy from a business continuity perspective.