Network Communications News (NCN) June 2016 | Page 46

COLUMN know how

On the right track

John Volanthen of Hub Network Services explains how to go about navigating the network jungle .

Moving IT infrastructure to one or more colocation data centres and putting it in the hands of industry experts can be a vital first step in ensuring your business is fully secure and protected . However , there is a danger of organisations overly focusing on data security either by hosting servers within a controlled data centre environment , or more commonly outsourcing to the cloud . This can result in the most appropriate network design and connectivity solutions being overlooked .

Having blown the budget on the best environment money can buy network managers and installers are often forced to make do with the cheapest connectivity they can find . This is usually much less diverse or resilient than perhaps it should be .
In order to avoid risking external networks becoming the weakest link in the chain , all the best practice taking place inside the data centre should be considered secondary to the careful evaluation and investment in stable , robust and truly diverse fibre links between locations .
Putting this into real life context in the outside world , networks are exposed day in day out to many risks , both environmental , such as frost , fire and flood , and man-made . Fibres can be prone to accidental damage not to mention vandalism which put matters beyond your control . Wheely bins set on fire and thrown down open manholes being one such example , time to fix
SLAs going AWOL when the fire brigade prevents access being another .
Therefore , to mitigate failure outside the data centre fortress it is essential to thoroughly evaluate which fibre routes and providers to use . These are complex to ‘ unravel ’ and external guidance may be advisable .
For optimised reliability , always connect via two or more independent fibre links , sourced from different carriers , with different physical routing . This tried and tested system optimises uptime and performance , giving the ability to transparently route traffic to the best path in real time .
Careful consideration must also be given to which network is right for your type of business based on the applications and systems you are expecting to run over it . Pick the wrong one for the job and at best you may be spending more than you need to while at worst you could be seriously compromising your business continuity .
Leased lines , for example , can provide guaranteed symmetrical connectivity to the Internet or to a data centre on a network . They are ideal if you have several offices or buildings spread across a local area ; require guaranteed Internet bandwidth ; need to transfer large quantities of data quickly and reliably for data back up and disaster recovery . There are a variety of speeds and options including copper or broadband back up and fully redundant lines .
Multiprotocol Label Switching ( MPLS ) to all intents and purposes is a private network for securely connecting any number of your remote locations using a highly redundant network . This is useful for communicating privately without special configuration over leased lines , xDSL or Fibre to the Cabinet ( FTTC ).
Data Centre Internet Access is a further option for consistent high performance and reliability . Suitable for all types of mission critical applications , including data replication , voice , video and web hosting , you will ideally connect to a network via two independent links , which provide diverse paths from each data centre .
Choosing two or more geo-diverse data centres with proven peering and transit both nationally and internationally will provide enhanced Internet performance and reliability as well as global reach . Ensure there are truly diverse routes for fibres to the cabinet rather than shared pipes or single tracks that will not be fully redundant .
Furthermore , it is imperative each site is incorporated into a functional ring system and hardware clusters have sufficient distance separating them to negate local complications . As highlighted by the recent flooding in the North of England , for those companies affected , a secondary cluster in the less sodden South of the country would have greatly reduced the risk of downtime .
A word on latency : Building your own or using a regional data centre for getting ‘ closer ’ to local customers does not guarantee applications will run faster and more responsively . This is because Internet routing does not precisely follow or overlay onto a map of the world . It ’ s rather like expecting the London Tube map to be a faithful physical representation of London – it ’ s not . So deciding to connect to a data centre somewhere on the planet without first consulting the Internet ’ s ‘ view ’ of the world is like sticking a pin in a map and hoping for the best .
It is therefore essential before connecting to data centres to evaluate the available network coverage and bandwidth in and around the areas concerned , and the connections going into and out of the proposed facility . These will also need careful monitoring following connection which can be a full time job and may best be outsourced to a connectivity specialist . One which can provide real time visibility into the optimum routing of global traffic and where best to place data centres on the Internet ’ s ‘ version ’ of the world map .
In summary , to ensure true network diversity , due diligence on the available connectivity options and costs involved should be done at the outset rather than as an afterthought . Don ’ t forget , outside the controlled environment of the data centre it ’ s a network jungle out there and one which requires constant monitoring .
46