I've been a network guy for a long time now. So as I say what I'm about to say, I think it will make some sense to you older guys (like me). Some of you seasoned veterans out there, it would be nice to get your input on this post.
I had this conversation today with my sales guy about sizing a core environment for a particular potential customer. He was asking me if a certain model of Brocade switch would work (ICX7450) for a core switch, instead of a pair of ICX6610s. At this point in the conversation, I'm already a little uncomfortable (technically speaking). My first question to my sales guy is "How many devices are on the network?" His response was "Not sure, Ill have to verify. But no more than 200." At this point, I automatically will say "No, the 7450s wont work in the core. Go with the 6610s."
As the conversation continued, we did get back into the verification conversation. I told him to verify how many devices they had on the network. If its 70 or less, go with the 7450s. If its more than 70, go with the 6610s. We also did have the 10gig uplink conversation, as in "How many 10gig uplinks do they need?" No more than eight would be appropriate for the 7450s (two of them), but if more was needed, go with the 6610s.
Back to the first part of this conversation. How did I concluded that the magic number of 70 devices was the breaking point for the 7450s being a core (if they didn't need more than 8 10gig uplinks)? I told my sales guy this: "Look man, I don't have a 'fact' to go on here. Its a 'feel'. I cant say that 70 devices is the magic device number for a core switch for the 7450s. Its just what I'm comfortable with, technically speaking." But, that comfort level comes with technical experiences. Both in Brocade and Cisco environments. What particular models will do and what they perform like. Again, its a 'feel' from experiences. Not because I 'know for a fact' in these cases. I realize there are other things to consider (like traffic patterns, etc), but generically speaking, as time goes by and your experiences increase, you just kindof know these things. Even without having the hard proof.
If you truly have a 'feel' for something, go with that feel. Sometimes that IS the best proof you have.
This is the retired Shane Killen personal blog, an IT technical blog about configs and topics related to the Network and Security Engineer working with Cisco, Brocade, Check Point, and Palo Alto and Sonicwall. I hope this blog serves you well. -- May The Lord bless you and keep you. May He shine His face upon you, and bring you peace.
Wednesday, November 30, 2016
The Network Guy: Facts And Feels
Subscribe to: Post Comments (Atom)
I've only been in IT for about 8 years, 1 as a network administrator. I can't wait to get to a point where I can use past experiences and "feels" for solutions.ReplyDelete
This is very true!ReplyDelete
Most of the others (non-technicals) just don't understand this. They have an idea, so it should be valid.
I came to realize, after working as a network engineer for a couple of years (which I've been doing for over 10 years now). It is very important to know your network. When you know your network well, then troubleshooting problems becomes easier. I don't mean to say that all problems are so easily solved. Though most can be. Without intimate knowledge of the network, it will take longer to troubleshoot any issues. Of course there are always exceptions!
My experience anyways!
Thanks, and have a nice day!
Not to mention when it comes to a core, you're sizing for at least 5 years...usually more like 7-8, sometimes 10. So would the 7450's work for today? Maybe. But you're not just sizing for today, and you're not just sizing for normal traffic patterns. We always use the traffic/roads example for networks, because everyone understands that. 90% of the time highway 65 into Birmingham would do just fine with 2 lanes. So why are there 4? For that 10% when there are extra cars on the road during rush hour traffic. Would a traffic engineer (or just a regular person) ever say "Look, that 10% isn't worth the cost, lets just let everyone sit in traffic 2 hours every day at 7am and 5pm" ? Of course not. You build for capacity and growth. Then when 4 years down the road the company merges with another company and doubles and amount of devices on the network, you don't have to go ask for more money like an amateur. You get to tell them of course it'll handle it, and by the way lets go ahead and go 10G to the edge since you planned for that 4 years ago as well. That's what separates the real engineers from the admins.ReplyDelete