40 and 100 Gigabit Ethernet

Today there is a lot of talk about 10 Gigabit Ethernet, 40GE, and even 100GE to support high speed networks.

Tags: Leviton
  • E-Mail
40 and 100 Gigabit Ethernet Asef Baddar, senior technical manager — Middle East & Africa, Leviton
By  Asef Baddar Published  September 23, 2013

Today there is a lot of talk about 10 Gigabit Ethernet, 40GE, and even 100GE to support high speed networks. This is of course fuelled by data centre need for speed and supported by the IEEE standards publication.

The new IEEE standard 802.3ba was ratified in 2010 and addresses requirements for 40/100G using parallel optics. To some degree 40G and 100G are being used today, as equipment manufacturers such as Cisco, Brocade, and Extreme have made it available. The majority of application is for 10G network speeds connecting the EDA to HDA and 40G or 100G from the HDA to the MDA core/aggregation. The move to 40G or 100G in the infrastructure is easy, causing almost no downtime. The key for that to happen is proper data centre design from day one. Full utilisation of fibre solution must be done to enable smooth migration to 40G or 100G in the future. This makes it easy for data centre owners to make the right decision when going to active and have a purely electronics decision.

The 40G SR4 technology uses a total of 8 fibres per channel. 100G SR10 technology uses a total of 20 fibres. This fact alone may be a significant barrier for 100G implementation between layer 2 and aggregation/core switches. The cost for the infrastructure would nearly double from a cable perspective. However, from a migration aspect, there is currently in working session within IEEE the emerging standard of ‘4X25’. This new standard is proposing the use of 8 strands for 100G capability (25Gig lanes). However, latest ‘news’ shows that distance will be reduced to below 70 metres for OM3 and 100 metres for OM4. I believe that this is why the new TIA 942-A standard has solidified its position of OM3 as a minimum and OM4 as the recommended medium for DC applications. Looking ahead, IEEE should have the 4x25 standard emerging over the next two years or so which will enable this same client to upgrade or refresh the active switches which will then be capable of implementing the new 4x25 technology for 100G I/O.

Keep in mind that by that time I believe that a much higher percentage of servers will be utilising 10G I/O or will be virtualised, both will require higher bandwidth from the aggregation point out. The client will be able to utilise the existing infrastructure for connectivity transmission. An important point is that the existing channel count that the client has designed for the 40G SR4 system will maintain the same channel count when implementing the 4x25 technology, hence providing cost effective migration for today’s 10G to 40G to 100G requirements.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code