Central data centres are connectivity relays, real marketplaces. They bring together almost all the players in the digital value chain. The challenge is therefore to know how to identify them, to be able to recognize them in order to open a PoP (Point Of Presence).
There are three types of data centers: hyperscale, edge and core. Usually organized in a loop, each one has a very specific role in the organization of an IT architecture. It is very common to see players hosting their application in a hyperscale, deploying their IT in an edge and ensuring an optimized network path by creating peering links in a core.
Central DCs have a very specific importance and are therefore becoming real performance levers that determine many infrastructure choices.
But how to identify them? The easiest way is to consult the referencing databases of network players such as PEERING DB and to search for the data centre with the largest number of members. If they have differentiating elements such as the number of members and network ports available, cushioned network equipment or an extremely wide choice of actors, it is in peering that their main attraction lies.
In a central DC everyone is on an equal footing: everyone shares data via a physical connection from point A to point B. Regardless of the nature of the interconnection: peering, direct interconnection or transit, I know that everyone is within cable distance of my rack. I will therefore benefit from clustering effect. And the effect is virtuous, the more actors a central data center brings together, the more interconnection there is, the cheaper it is.
Peering is a strong trend that is becoming essential. A study published by Arcep in 2017 related to traffic measurement among ISPs in France indicates that the data exchanged on the territory are distributed in this way: 50% for transit, 46% for private peering and 4% for public peering. The same ratios were observed by the Journal du Net in one of its central data centers. The share of transit decreases very significantly between 2017 and 2018. Public peering is growing and private peering is increasing very significantly. Three main consequences follow from this dynamic: content players will get as close as possible to end customers by bypassing hosters and forwarders in the short and medium term, freight forwarders seeing their business decline will try to recover the margins they are losing on the CDN link, and finally ISPs will try to get closer to the end customer themselves by including content in their offers.
Several good practices deserve to be shared to move to live traffic. First, start with the application. Before choosing where to host your IT, it is necessary to consider the nature of the IT. Depending on the answer, you have to organize your architecture. The challenge is to create network accesses that facilitate the user experience and reduce costs. Depending on their priority and the level of security required, the applications will therefore be divided between core, edge and hyperscale.
Secondly, how to bring the user closer to these applications? The alternative is quite simple: either use peering or direct interconnections, or put the application locally in its data center and set up a private network link to the end user.
The meaning of the story seems to be moving towards a transformation of the IT agent into a buyer. IT managers are now able to organize these outsourcing choices in these three types of data centers. Business choices therefore become business choices.
Read the article
Source : Journal du Net
