Is the Cisco Nexus 7000 Needed Today — or Tomorrow ?

No doubt that the Cisco Nexus 7000 switch is a fine piece of tech­no­logy. The per­form­ance and through­put is wel­come, and clearly offers some fine new cap­ab­il­it­ies such as vir­tu­al­isa­tion, ISSU, bet­ter OOB and so on. I am sure that every­one can per­ceive the pos­it­ive mes­sages, lets face it, Cisco isn’t going to be shy in telling us about them.

However, lets con­sider the issue from the per­spect­ive of the architect/​designer and how Cisco has posi­tioned this in the mar­ket­place. From an archi­tec­ture per­spect­ive, I will need to com­mit a sub­stan­tial capex to the product and a much lar­ger amount of resource cost to trans­ition a net­work to use the product. Even if I am build­ing new data centres (and thus have no leg­acy), changes to oper­at­ing stand­ards, pro­ced­ures, man­age­ment tools and other orches­tra­tion issues present sub­stan­tial bar­ri­ers to adoption.

New OS, New technology

It is a given that this product has been extens­ively tested, and will work as described, prob­ably with a small num­ber of the usual glitches. Cisco com­mit­ment to soft­ware test­ing that com­menced around 2004 /​ 2005 has def­in­itely reduced the num­ber of sig­ni­fic­ant prob­lems in the field. I do not believe that we will see repeat of the IOS 12.1.5T where so many fea­tures were intro­duced and took years to finally sta­bil­ise. Notwithstanding, NXOS is new and needs care­ful con­sid­er­a­tion. Currently, I am most con­cerned about integ­rat­ing NXOS into my cur­rent NMM plat­forms. I sus­pect that Cisco will have addressed this at some level by delib­er­ately main­tain­ing SNMP MIBs and the IOS CLI but NXOS will have dif­fer­ent con­sid­er­a­tions for CPU, memory, and other key mon­it­or­ing vari­ables that will take some time to adopt.

Fibre Channel Over Ethernet (FCoE)

The move to integ­rate FCoE into the Supervisor is not entirely unex­pec­ted. It is in Cisco’s interest to lever­age their strength in Ethernet switch­ing to extend the mar­ket share of the SAN mar­ket. However, what is strik­ing is that Cisco has delivered a FCoE solu­tion before the stand­ard is com­pleted. But Cisco has made this a key part of their mar­ket­ing message.

Cisco is a reg­u­lar player in the stand­ards bod­ies, and we have seen Cisco deliver tech­no­lo­gies such as HSRP and ISL, long before IETF /​ ANSI /​ etc have even star­ted stand­ards pro­cess. The key dif­fer­ence here is that no inter­op­er­ab­il­ity was ever needed. Thus HSRP only worked on Cisco Catalyst switches.

FCoE requires to co-​​operation of com­pan­ies who are notori­ously closed organ­isa­tions. Indeed, Fibre Channel SAN can be viewed as a move by Storage vendors to make sure that they owned the entire solu­tion to max­im­ise their con­trol of the cus­tomer (prob­ably to the customer’s cost). Is it reas­on­able to expect that EMC and NetApps are will­ing to forgo the rev­enue of Fibre chan­nel adapters and switches ?

Approvals and Validation

One of the greatest mys­ter­ies to me, is how Storage Vendors have con­vinced their cus­tom­ers that val­id­a­tion is a pre­requis­ite for sup­port. This seems an archaic concept hark­ing back to the hal­cyon days of main­frame, and def­in­itely not for the better.

Every stor­age equip­ment pur­chase is reg­u­larly vet­ted by the vendor (or worse, vendors) and long lists of “approved equip­ment” are con­sul­ted prior to any pur­chase. And yet, in net­work­ing, we have no require­ment for this pro­cess. Our stand­ard are clearly defined, and the mar­ket quickly dis­ab­uses vendors from intro­du­cing incom­pat­ible features.

Quite why test­ing and val­id­a­tion takes months is also a mys­tery to me. The acquis­i­tion of VMware by EMC has seen a marked rise in this prob­lem. Not too long ago, VMware was used on any equip­ment, and sup­por­ted by VMware.

Is Cisco going to gain these approvals in time for us to use Fibrechannel ? Are EMC /​ NetApps /​ VMware going to sup­port them ? Also, will there be qual­ity drivers for Windows Server 2008 and 2003 ? These unknowns are a con­cern to me.

Networking versus Storage

For very large data centres, the Nexus will take a long time to reach accept­ance. The mer­ger of Storage and Networking people into a single team will take some time due to human res­ist­ance to change, and the the changes to oper­a­tional pro­ced­ures will take a long time to achieve.

Storage people will have a lot of per­sonal invest­ment in stor­age products both in terms of exper­i­ence and cer­ti­fic­a­tion, and pro­fes­sional pride. Also, they are res­ist­ant to change, some pock­ets are deeply hos­tile to new advances.

The crit­ical nature of stor­age means slow take up of the new. Comments such “we trust EMC” or “NetApps is proven reli­able” and “com­pli­ant” and “cer­ti­fied” are often heard.

While I agree that this has happened before with TDM Telephony to VoIP, this is no guar­an­tee that the con­ver­gence of stor­age and net­work is a fore­gone conclusion.

Nothing to Lose

It is worth not­ing, that due to the archi­tec­ture of the NXOS a a fully vir­tu­al­ised core, that if FCoE doesn’t gain accept­ance, then Cisco can just remove it from the soft­ware. The soft­ware will still be a mon­ster Ethernet switch plat­form and Cisco will only lose the invest­ment in FCoE not the entire platform.

Penetrating the Server Farms

Still this rep­res­ents a major shift for Cisco, which has struggled to make its pres­ence felt in the Storage space (although they claim to have 50% of the space that VSAN space that MDS occu­pies). Since SAN are largely the pre­serve of the server teams, Cisco — as a net­work com­pany — may have dif­fi­culty gain­ing mind­share. I note that Intel has put a lot of weight into FCoE, so Cisco is not alone here.

Early Upgrades — how many ?

The first ver­sions of the Nexus 7000 have a capa­city of 230Gbps per slot, but the total slot per­form­ance is an order of mag­nitude lar­ger than that. Logically this means that Supervisor and Line Card upgrades are going to be part of the life­cycle to reach to higher performance.

This isn’t new. People who pur­chased the first Catalyst 6000 units have had eight or nine upgrade cycles to reach the max­imum capa­city of the chassis. In 1999, the first Supervisor 1 sup­por­ted switch­ing only and had a 32 gig­abit per second shared back­plane. It took a few years before we had the fab­ric switch­ing mod­ules and rout­ing super­visors (using fea­ture cards) and five or six years until we got to the 20 Gbps per slot capa­city that we have today, and the first high speed mod­ules were really expensive.

The most import­ant fact that I remem­ber, was the first two or three years was an end­less cycle of upgrades to get the basic fea­tures that I wanted. As an old (not bold) engin­eer, I am wary of enter­ing this cycle again.

Missing Features — so many.…

There are so many fea­tures miss­ing in the product. MPLS is one the biggest, you need to look really care­fully to real­ise that the cur­rent ver­sion does almost noth­ing but switch­ing and the most basic rout­ing. Once you real­ise this, you can only pos­i­tion the NX7000 as either a high speed eth­er­net switch in your data cen­ter core for ter­min­at­ing 10GE, or as an access switch for serv­ers that are using 10GE where, hope­fully, you might even be using FCoE. Spending sev­eral hun­dred thou­sand dol­lars on an access switch is going to be difficult.

Splash the Marketplace

If you accept my pre­vi­ous com­ments, then I believe has Cisco released the NX7000 fol­low­ing reasons:

• to head off the entry of other 10GE vendors such as Juniper, Woven and Foundry to their cus­tom­ers. ( This also will scare off fur­ther ven­ture cap­ital fund­ing on 10GE )
• to put the product into the design pipeline of cor­por­ates and ser­vice pro­viders, in the expect­a­tion that deploy­ment will prob­ably start in two to three years, by which time the fea­tures have been delivered.
• to show com­mit­ment to FCoE and provide mar­ket­ing impetus. Some cus­tom­ers will assume that if Cisco does it, its going to be big (and most likely they will be right).
• to shake up the stor­age mar­ket­place, and get the cus­tom­ers ask­ing the Storage com­pan­ies to release FCoE products, with approvals!
• because the C6500 is show­ing grey hairs. The recent VSS upgrades don’t improve per­form­ance, just extend the life of the cur­rent solu­tion. More upgrades!
• Fiber chan­nel adapters and switches con­sume extra­vag­ant amounts of power( a big weak point). By lay­ing out the FCoE com­mit­ment, fund­ing for new hard­ware pro­grams on FC for power reduc­tion will be less likely.

Conclusions

It is prob­ably too early for most of us to rush out and pro­cure the Nexus 7000, it costs too much, it needs time for man­age­ment tools to mature, and it isn’t a ‘must have’ right now. I sus­pect Cisco is mark­ing out ter­rit­ory to pre­vent com­pet­i­tion. I am uncon­vinced the FCoE will change the world, and con­cerned that many fea­tures are missing.

I think the product needs quite some time to mature before I will be ready to take the plunge.

Please rate this post:

(No Ratings Yet)

 Loading ...