... we've filed patents
Does anyone else see flashing lights and hear wailing sirens?
If you are a network administrator, be aware that there are a lot of industry movers and shakers who want to put you out of a job. Mike Banic, marketing VP for HP's networking division, exemplified that goal when speaking on a panel at the Ethernet Innovation Summit, held this week in Mountain View, California, to celebrate …
How many network admins will even look at HP products, especially now?
What he is also forgetting is that when there is an issue, software won't fix it. I can't even count how many times there has been an application performance issue. It can be network, server or application or all combined. Software cannot fix those issues as it is all a manual process.
Certainly, there are many reasons to not consider HP products (SDN potential being #2 on my list after their ridiculous command-line nomenclature), but HP isn't going it alone in pushing SDN. Expect to see every player in the space moving this direction, if they aren't already.
The more important issue that no one seems to want to address is, "What happens when it all goes titsup?" Someone will damn well wish they had kept some engineers around for that day.
So? IT has put a lot of non-IT people out of work, with IT people getting jobs supervising/implementing/whatever that automation. Arguably this is just more of the same - karma, poetic justice, call it what you will. Look on the bright side, though - if/when things go badly wrong you can be a consultant!
I fail to find myself trembling in my shoes. I've seen first-hand what happens when a certain variety of management monkey imagines they know enough to fire the help and save some cash. Actually my sides still hurt a little sometimes.
I have a job working back at a firm where I set up the IT along with a couple of colleagues. They laid us off and did IT with non technical staff and used a support company where they couldn't handle things.
I'm back working there because apparently nobody realised that a drive failing in a RAID array was bad, and feeding tapes into the server every day was boring and old fashioned. I'm given to understand that this attitude changed quickly after the failure of a second drive in said RAID array.
to see if this comes true or not.. everything I see tells me this mass network automation in SDN really only applies to massive scale deployments(service providers). It's not like say storage automation, which is useful at all levels of scale(even when that level is a single unit acting alone).
I have a friend who told me a while back that (at least at the time) Facebook's entire network was run by something like 7 people. So if you design things in a way that it's scalable, and easy to use then you can certainly reduce the # of engineers.
The more important task will be ease of use. Going from 1 network engineer to 0 is a big step. That is going from having dedicated trained people where most of what they do is sit around waiting for something to break to transferring that responsibility to people where that is not their primary role.
Most network(switch/router) equipment user experiences are very poor and not friendly. This is "OK" if that is your full time job you deal with it all the time. Less OK if it's not. Load balancers (many say these are still switches just operating at Layer 7) often have very user friendly UIs and are easy to manage by contrast. Even many firewalls have relatively easy to use UI vs switches/routers (there are exceptions in both cases).
There will certainly be push back from those whose jobs are endangered, it is unfortunate in many respects to see how automation continues to really go to permanently lower the # of workers required in the workforce. I think back in time, the stuff I do today, vs how it was done a decade ago and how many extra bodies would of been required, how I could double or even triple what I do and still be OK. A decade ago I was in a similar situation though things were much more chaotic, costs were significantly higher, many more people, many more problems. Today things hum. Mainly due to automation (& consolidation) at various levels. My job security is not in doubt for a while but I am alarmed at the trend.
When you can condense 30 racks of equipment down to 3, you don't need as many people to manage that. I had precisely that proposal at one company a few years ago but their short term budget priorities were elsewhere (company is almost out of business now). My former boss at the company told me recently the company hired upwards of 9 people to replace me after I left(technically to replace a team of 3 - the other 2 on my team I could of done their jobs in in my sleep on top of my own). Oh that was so hilarious.
In the storage realm a friend of mine told me a story he encountered while trying to pitch 3PAR at a large EMC customer many years ago. The room was filled with storage admins, and they were all very impressed. The manager/director told my friend the technology looks awesome - come back in 24 months and we can make a deal I'm sure. My friend was confused -- what's with 24 months? The customer said - I'm retiring in 24 months, and if I brought your stuff in today I would not have the need for all of the storage admins I have now, I would have to let some of them go. I feel I must try to defend their jobs while I can, so that means we need to stick to EMC to support our staffing levels.
I can certainly appreciate the loyalty a boss would have to their staff, and at least personally I can totally understand the decision the customer made at the time.
It's sad to me to see how we as a civilization are not able to find alternative means for these displaced workers to do stuff at a fast enough rate(even Foxconn is planning to replace hundreds of thousands of workers with robots soon). This doesn't bode well for anyone.
The race to the bottom continues, it's a race most of us (myself included) are likely to lose
The interface issue is one reason why Juniper and Netgear are making big inroads on the old iron, like HP and Cisco. The downside is that their really hot tech is still very pricey and out of reach of any but the most dedicated midrange business, whereas just buying one more HP is cheaper in the short run, even if it's much less powerful. Since salaries come from a different budget, there will always be a conflict.
Don't get me wrong, I love me some CLI, but Netgear (and I presume many other) "Smart" or "Managed" SME switches have the ability to simply back up their running configs over HTTPS. In fact, I have some 52 port 10/1Gbps Netgear switches that actually have 2 NVRAM pools. You put your default stack configuration in NVRAM1, and then back it up to NVRAM2 after confirming everything works (make sure to download that old backup into cold storage, maybe a USB drive, first). If you screw something up, just change the running config over to NVRAM2 and you're fine. The configuration propagates to all switches in the stack in less than 90 seconds. I doubt you could even ssh/telnet/walk over to and plug in a console cable/etc to that many switches and log in (let alone paste and activate your configs) given the same time period.
If you screw up royally, or a rampaging escaped gorilla destroys one or more switches, you can just upload a cold storage backup via the web interface. Again, those settings propagate to all stacked switches in less than 90 seconds. If you feel like it (just hate the GUI) you can even open the backup file in your favorite text editor and make any changes you like there. You still have to know what you're doing, but I don't miss Cisco's IOS at all. You can't "copy run start1" and "start2", let alone set up link aggregation groups involving ports on multiple switches using their hugely more expensive hardware. When a $1,300 Netgear outclasses a $4,000 Cisco you have to wonder.
The point about dual NVRAM mount points still stands and mLAG != LACP. The switches themselves obviously know what's going on, but they're not going to appear as a single logical switch to attached devices which breaks LACP. Therefore, with mLAG your switches are making decisions independent of the LACP settings on attached servers. STP would probably still bail you out if you did something dumb enough (oops wrong port), but having no automatic verification sounds frightening. I really don't think Cisco/HP have anything worth looking at that's less than $100/port, which is the real point I was trying to make.
Saying the 6500 series is for SME is kind of a stretch in this economy (though VSS is easily as good or better than anything I've seen from Netgear/Dlink/ect.). I mean those are like $4,000 for the damn chassis. You still need to buy modules, fans, and power supplies on top of that. Got multiple buildings? Then you'll need a chassis in each one. Good luck getting that approved.
You're right to point out Cisco Nexus gear as vastly superior, but it's just completely out of the question financially so it'd better be. If your SME can afford those then you're going to hire a koolade drinking CCNP or better anyway so Netgear/Dlink/HP/Dell/etc are out. I'd argue that Juniper would be the lower price gear to compare to those, but most enterprise guys I know act like Juniper doesn't even exist. I'm admittedly in over my head on "real" Enterprise gear, so perhaps there's a good reason (other than that they've spent their entire careers working in the Cisco ecosystem) that they seem to dismiss Juniper.
Given the dismal failure of the network management software industry (I'm looking at you, HP OpenView) to solve even basic network management problems, I doubt they will ever deliver on the vision. They will, however, be successful at selling boatloads of overpriced shelfware to clueless & gullible IT managers.
Sadly, those clueless managers will continue to believe the hype and invest heavily in something which doesn't work. Just like the last time. And the time before that. And then, when their new automated empires come crashing down around them, they'll suddenly realise they're fresh out of options and throw cash at those skillful enough to recover the situation. And in about five or six years, when they've been promoted, and a fresh-faced, inexperienced team come in to replace them? We'll see the whole thing kick off all over again!
So you're trying to manage the business critical network infrastructure of a major organisation, and at the same time please your line manager by saving some money through firing network administrators who have a bit of experience and know what they're doing.
You do this by reading "HP is working on automated virtual application networks that treat a network as a unified pool of resources rather than as an agglomeration of individual devices" and thinking 'Gosh, that sounds impressive - I'll bet the farm on that one".
Well, good luck with that. The next time that a problem occurs which isn't covered by the "automated virtual application network's" set of problem solving menu screens, who you gonna call? GhostBusters?
Sometimes I despair.
...Not with my Novell Network skills I'm not! Seriously though, we're in an era of cost cutting galore. Salesmen selling Network Management Software or the Cloud will undoubtedly manage to persuade some management teams that its in their best interest to oust some Network / Sys Admins. In a time of rising security problems this can only lead to major disaster IMHO.
Ahhh...talented Network Admins...
System upgrades or failures are the only times the importance of your existence is appreciated or acknowledged.
Otherwise you're an overpriced obstacle in head count reduction that can be replaced by a rent-a-tech whose communications skills and personal hygiene are both abysmal or a shiny black box. Or so the vendor's Powepoint presentation would have you believe.
I survive thanks to the ever-increasing demand for my Banyan Vines 4.11 Network Administration certification.
And the clock just rang Beer O'Clock as I finish typing this.
Technology has always been about reducing the need for human beings. One needs fewer people to conquer another if one group has bullets and the other has arrows or one has metal while the other has only wood.
No news here.
Everybody likes to think what they do is so important to someone or something. Not true. We are all dispensable, it's just a question of how and when.
If you work in technology, you are responsible for putting people out of work. That is our job. Eventually, they come for us.
There is nothing evil in this. Adapt or die is the foundational principle around which all life operates no matter how advanced.
It's a process one has to endure while one is alive but the good news is, it's not forever.
We all know this. We just don't like it. We still think of ourselves in small, tribal hunter-gatherer terms. Such thinking became obsolete as soon as we banded together into larger collectives. Today, the human being is worth less than ever. We are components of systems that operate independently of us. This is by far a minority view but I see little evidence to the contrary.
The more ironic take from this vein of thinking is that business/MBA types are the source of all this optimization, when in fact it's really been techies that have driven the move to automation and process standardization. Business types aren't the unbelievable evil that everyone thinks of them, unless you're willing to throw techies under the same bus.
That isn't always the case it is most of the time though.
The truly indispensable people (Usually who have done something like written a legacy billing system).
Often end up getting paid more as a contractor to sort of the mess that was totally unnecessary in the first place.
"If you work in technology, you are responsible for putting people out of work. That is our job."
No need to buy into the Luddite analysis. One could equally make the case that we are responsible for keeping people in work, by raising their productivity to the point where it can finance the salary that pays the bills charged by the rest of society. There are more people "in work" these days than at any point in human history.
As a long-term EvE Online player, 'Adapt or Die' is a personal mantra now. I'll adapt. It's the same thing that set off the Luddite movement, now we're facing the fact that as technology improves, certain skills will go by the wayside, and a neo-techno luddite movement could spawn.
The way to survive it is to learn new skills.
As someone who has worked with automation most of my working life, I can tell you it is not and never will be thanks to the same cheap ass sentiment toward QC as they have to everything else just in general principle.
As someone who now works for large world domination computer company, my job is to fix the things can never be automated, i.e. users connecting to the network.
Sometimes it's the users, most of the time it's the equipment and software on both ends.
But guess what? It's going to happen anyway because the entire workaday part of the industry is staffed by Randians who think they are far too clever individually to work together for their own good.
People are beginning to realise most of the high end (read Cisco IOS console-cable) 'skills' are smoke and mirrors. .A bit of basic networking knowledge and picking a product that isn't overpriced, buggy, semi-proprietary and needlessly obscure will solve most companies' needs. Let the downvotes commence.
I don't get it. If you just want a pool of stuff, you can have a flat network (all switched), DHCP, and printers and etc. all support autodiscovery (in Windows, OSX, and Linux at least), file shares (windows shared, nfs, iSCSI) allow disks over the network, and it's all is just there. If you're looking for a pool of processors, there's "cloud" distros that will just keep netbooting machines as they are plugged in. But, once you want anything more complicated than this then...? Well, that's when you need some humans in the loop to at least make actual policy decisions.
Also, how does an Exchange server setup take 2 weeks to set up? I mean, any UNIX mail setup I've ever seen, you throw together enough RAM, CPUs, and disks to handle the load, load the mail software on there and start adding accounts.
Actually it probably takes an hour or two to assemble the kit (or let the CE do it), power it up, test it, load
the software and test it, etc.
The other 13 days 22-23 hours are taken with Change Requests/Change
Management, coordinating everthing with/from India/Malaysia/CostaRica/take your pick (hey, eventually
we'll be so broke and downtrodden they'll eventually outsource back to us.....), then the misbegotten
marketing misanthropes will pat themselves on the back while "Rome" burns.....
I guess the folks from Golgafrinchan ship B got tired at looking at moodily lit tubes of toothpaste....
gets brought up in a meeting, I can see the network specialist change into snooze mode "Not part of my role, my job is waaaaay too complex to reduce with efficiency like this". I would rate networks as the 2nd most ignorant of the IT disciplines about how other disciplines fulfill their function, and the more senior the tech, the more they seem to revel in this ignorance. Any measures designed to bring them to heel are welcome and should be encouraged to the n'th degree.
I've got a theory on this - it's because many network admins started out running cables. They then moved to the console, and can configure what they need. But they haven't been exposed to basic programming or scripting.
I think that we're going to see a big shift in the way networks are managed in the next few years, and there will be far more automation, and far more centralised control of networks. People with Unix admin or programming backgrounds will adapt quickly to it, but it will be very tough for those without the coding background.
Maybe what we'll see is things like VMware/Nicira gain widespread adoption, but it won't be managed by the network admins - it will be managed by the infrastructure teams. The network admins will just provide basic connectivity between servers, and the more interesting networking will be managed from vCenter. Obviously that doesn't cover all the complexities. Will be interesting to see how it turns out.
actually (and i've seen this in many other places too), we, the networking people, provide the networking links to the chassis of the VM systems (in our case ESXi hosts on Cisco UCS chassis) that run the Nexus 1000V software switch. the systems people can now do ALL their stuff, spinning up new servers etc on the 'correct VLAN' without needing to touch us at all - we provide all the routing, VLANs etc, we provide the name of the VLAN via the nexus..and they just get on with it....which means that WE can just get on with the other stuff such as dealing with the rest of the enterprise network, keeping up with technology etc. I see OpenStack, I see software and virtual switches becoming more common place....free me from mundane network admin stuff and I'll have the time to get this automation stuff into place so that we can ditch old methods and go with new ways (thinking spanning-tree to TRILL, OSPF to ISIS (we've done that), OVT etc). we've already deployed IPv6 , DNSSEC etc because we use automation to a degree. but other posters are right...a PROPER network - ie not just some nasty flat space, needs context, policy decisions etc for defining what goes where, who does what. 802.1X itself needs policies and changes... the important part is AGILITY, not to be confused with SPEED. (and perhaps some places are paying their network admins too much? I certainly dont get the salaries that are banded around in the press!)
Yes, we all know the end-game is to eliminate the meat-bag. Nature of the beast and no news there.
What I resent, however, is being told this by people in MARKETING! How come these bottom-feeders are allowed to have 'a salary that goes up with inflation every year' - why won't GoogleAds make you redundant? Or better still some sort collective epiphany that leads you all to suddenly realise the pointless vacuity of your existence and makes you do a proper, useful job instead that involves helping people or actually making stuff.
The guy probably is some marketing cone-head left over from the Carly era...."perceived needs" and all that...or "Perception is reality".....instead of a burning platform, the round burning toilet seat of Lucent came to HP...
Sad how the mighty HP has fallen.....they used to drive innovation (and didn't need a stupid "INVENT" tagged on the bottom of the corporate logo in order to know to do that)...and they created markets rather than trying to convince people that they needed something useless....
Most of what they have now is hideously overpriced ink...no true innovation there except for folks who consider money or, more often, e-cash changing hands as a "product"....BAH....
While the marketing douche provided the fodder, don't believe for a second that this originated from their marketing abyss. Network automation is being pushed from the technical/innovation side of the house. I think it's doomed to fail myself, but you should really find a new story to overly bemoan than "HP innovation is dead" and "it's so sad". Your inability to overcome your own personal disappointments makes you far sadder than you realize. Stop mourning and get on with your life.
Which ho did you save? EDS? DEC? Compq? Apollo? The various medical and instrument companies HP absorbed in the 50s and 60s?
I'm doing quite fine, thank you, since I left HP....and I suspect you never worked there (at least when it innovated)....let me know when you get those self-installing cables/switches/routers/etc. installed that can take care of obstinate users/managers/etc. as well....
I am sick and tired of some dBase relic sitting in an ivory tower with a VAX in the corner shouting achtung! whenever he wants, and making all of us goose-step to his orders. Foisting the likes of Lotus Notes and other outdated garbage on innocents should be criminal.
IT has a real PR problem with the rest of the employees in their companies, and they are oblivious to it.
It's not we, the masses, who need to change, it is they. Computing is not what it used to be, and the sooner they hit the soup lines the better. They need to wise up before that happens. And it WILL happen.
Achtung? Ex-Compaq (or DEC) are we? Or perhaps marketing, as you're so worried about PR versus actual work and results ("Perception is reality" according to Carly)....
Er, right, you can lash up a mail network on your iDevice and pretend it's a real exchange or other large-scale mail server environment...
you seem to be confusing a network administrator whose main focus in life it to help IT actually work (with the first imperative to be that the network remains safe and stable!) with some Operator (eg BOFH) or senior IT director.
A network admin lives in layer 1 - layer 4 most of the time. we dont CARE what email app you choose to run or email platform you use, or groupware. we DO care that it follows specifications and that theres clear documentation about what protocols/ports/etc it needs to run as we need to ensure that things like firewalls and ACLs are sorted out. we care that its not a pile of crock like microsofts load balancing sh*t. I embrace things like BYOD (and have done for much longer than the recent industry hype about it!) - but someone else writes policies on data retention/control etc.
..and anyway...who're you going to blame when things take time or go wrong if its all just some automagic HP software stack (and based on their (and ciscos) historic and current coding and software delivery it IS going to go wrong)
Haven't you heard? it's all part of the new software defined network. You just get anyone off the street to knock up the racking, bung in the servers, switches etc. and stick all the plugs into all the sockets until there are no holes left.
The system is so clever it can work out where they all go (even the ones that come out of one port and go back into another port of the same switch) and optimise all the connections.
I remember some dweeb from HP saying the same thing about SysAdmins a few years back. I seem to recall the pitch was about HPSIM. It would auto-discover new servers, pick up faults, alert HP automatically when any fault cropped up.... apparently it would reduce the number of sysadmins required, and the ones left could spend more time in the pub!
But when you look at it its based on SNMP, and when SNMP isn't working it defaults to 'all OK' even when the server is riddled with faults. Coupled with a crap database design and alerting system on the backend its basically unusable.
I have no doubt any networking software they produce will look and behave the same.....
"SNMP isn't working it defaults to 'all OK' " Sometimes I misread it as "SMNP" (need new visual hardware, I suspect)... Simple-Minded Network Protocol .... probably what they use/need here...
HP has been vacillating between hardware and software focus for at least 30 years..too bad they've given up so many times....once upon a time they had some good OSs and some good applications software, but never seemed to commit to the idea they could do anything other than hardware (or the limited software to support it)...Now most of their hardware is just off-the-shelf crud with nothing to distinguish it from all the other white boxes out there....
These "industry execs" need to quite drinking their own koolaid. Coming from vendors that innovate with 2-year HW lifecycles. Policy application and management will improve but how the evolution will look is most likely similar to the x86 virtualization. Juniper and HP should probably be focusing on real solutions rather then pondering about 10-30 year future of network ops.
I would put money on those companies being completely irrelevant in 10 years before I would bet on network engineers to be irrelevant. They should probably go have a field trip on how the Internet works today and how at risk their hardware portfolios are today.
The reason it might take two weeks to deploy an Exchange server in an organisation is the administrative overhead of getting approvals to do the work, not actually doing the work!
When business process demands that requests get routed through a variety of little Hitlers who can derail the process at any moment, you can end up spending a lot of time crafting a request with an appropriate amount of grease to allow it to slip through unimpeded.
Error checking, quality control and dealing with political fallout are all tools left to human beans, unless we're eliminating them at the same time?
The comments have generated enough meat for the crock pot, so bring on the rest.
Or just dump the whole thing down the disposal, including the IT types. We don't need them, want them, nor will we continue to tolerate them impeding business.
They just don't get it...they really, really don't. Business and productivity are not in their lexicon. Paycheck is, and that's about to change.
"How many network admins does it take to change a light bulb?"
Answer1: None. We can code in the dark.
Answer2: None. That's a hardware issue.
Answer3: WTF? There is no way that my code could have killed an AC lightbulb.
Answer4: None. Simply call facilities.
Answer5: None. I have a Sun on/under my desk.
Answer6: Light bulb? That reminds me, I just had an idea ...
Answer7: OH FUCKING CRAP, some asshole just ::snipped:: the building neutral at the fuse panel, thus tying L1 and L2 together ...
I got a million of 'em ...
I do think that we will see more devops type roles. But I don't know that it spells doom for all network engineers. I think of this shift a lot like the ERP shift. Sure, you have people who architect across silos, but you still require people with specific systems expertise, particularly when things go wrong. The higher-level architectural type tasks will elevate above specialists, but at the end of the day, you still need someone who knows how this stuff works.
If people essentially let go of all of their specialists, they run the risk of not having them when they need them. You don't need someone who knows how your carburetor works... until you do. If you can afford to take your network to the mechanic shop and wait, then that's fine. If not, then you need to have someone who can cover deeply technical issues as needed.
He who builds the entity which "builds" the automated virtual application network that treats all networks as a unified pool of personal and exclusive resources, rather than as an agglomeration of individual devices for global domination and zeroday vulnerability exploitation of Cyber Command and Control and SCADA Systems with other sociopathic mentality players, will be the smartest guy in the room, and in the known universe of global operating devices spinning planetary phorms which creates Future Reality Programs and Projects with AIdDefinite Vision.
And the entity will be able to destroy anything it desires and all that might eventually learn to think to oppose or compete against its network allies and/or affiliate virtual applications. Imagine, and it's true, and it and IT be for Real whenever Virtual in Play …. and that be AI@ITsWork for triumphs online in Live Operational Virtual Environments/Immaculate Reality Applications
When it comes to the fate of network architects and engineers, I respectfully disagree with my industry colleagues and actually see the move to network virtualization and SDN as an opportunity for the networking team. I’ve shared my thinking about this on the Embrane blog and welcome the continued discussion.
Biting the hand that feeds IT © 1998–2021