While there seem to be a plethora of stories about how "cloud" (private or public) computing is and will revolutionize the way we work and how products are brought to market, has this really affected the way that technical people do their jobs or find new work?
For me, the biggest changes have come in how the systems I'm responsible for are deployed, secured and managed. But aside from changes in some management tools and implementation scenarios, things haven't changed all that much.
So how has the advent of more centralized computing (resource pooling, application/workload partitioning, fluid resource allocation, etc.) affected how and with whom you do your work?
This discussion has been archived.
No new comments can be posted.
Has "Cloud" Computing Affected How and/or With Whom you Bring your Skills to Market?
|
Log In/Create an Account
| Top
| 21 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(Score: 1) by Buck Feta on Tuesday November 25 2014, @02:49PM
I bought some water at 7-11 yesterday that was both gluten free and cloud enabled.
- fractious political commentary goes here -
(Score: 3, Funny) by WizardFusion on Tuesday November 25 2014, @03:29PM
Make sure your intake of Dihydrogen Monoxide is limited - that stuff will kill you in large doses.
(Score: 2) by davester666 on Tuesday November 25 2014, @06:36PM
blech. it was loaded with sodium cloride.
(Score: 4, Insightful) by VLM on Tuesday November 25 2014, @02:53PM
things haven't changed all that much.
I used to do maybe 5 hours a year of hardware stuff. Reliable gear doesn't need much Fing with it. Once in awhile a drive would die in a RAID and I'd replace it. Believe it or not, its been decades since I've restored data from backups although I still keep them. You can't F around with RAID, mirror the heck out of it and never use two drives from the same mfgr or sequential serial numbers and put each drive on a different controller and do software RAID only and you'll be OK.
Now I spend maybe 40-60 hours a year on brokering boxing matches between the vmware guys and the NAS guys who apparently hate each other and based on performance problems only interconnect via an old piece of 10 meg thinnet, or maybe arcnet. And complaining when I lose vsphere client access. And when they move my images without telling me (because its supposed to be transparent no impact) they inevitably lock something up about 1 time in 10. Because IT reorgs more often than I interact with them, its a true pleasure to try and figure out who to contact every time. And there are no written standards because it takes longer to develop them than to reorg the department (seriously, bad seagull manager problems over there.)
So I'd say I spend about 10 times as much time on virtualized hardware problems as I did on actual hardware problems. Its strangely relaxing not being able to fix anything anymore. Less work to do, but stuff doesn't work as well. On the other hand it makes me look bad. I shut down for two hours to replace the oldest drive just based on age and blow the dust out, the users can understand because they saw me doing physical work. The NAS is barfing again providing floppy drive level sustained thruput (50 k/s or so) and the users see me lounging at my desk because what else can I do? Stuff doesn't work? Well duh, clouds don't work!
This is all private cloud internal only, public clouds might be more reliable. How could they be worse and still stay in business?
(Score: 2) by tynin on Tuesday November 25 2014, @03:17PM
You said everything I wanted to say.
My story is I touch ~9k physical servers with maybe ~60k HDD's. We lose a drive or two a day, no big deal everything is RAID'd, most everything also has N+1 replication of important data that autocorrects in the event of total server loss (and often even for full chassis loss). Things just work, the system can be a real joy to support.
On the cloud side of things, I deal with 108 physical cloud servers with maybe ~1k VMs. By far and away the majority of my day is spent fixing crap in the cloud. Single problems cascade into dozens if not hundreds of faults. Cleanup often takes weeks when something as basic as a server going offline occurs. The system got too big too quick and of course we put production services on it right away, so it turned into a frankenstein that cannot be properly fixed short of a rebuild. So we stay in a constant state of disrepair with one offs and bandaids the norm. Management seemingly doesn't care at how ridiculously inefficient the whole mess is. Hopefully we don't grow it, and in a few years put it out of misery.
(Score: 3) by WillR on Tuesday November 25 2014, @05:01PM
Business alchemy! They transmogrify capital expenditures into fee-for-service. Buying a new server needs sign-offs from 5 levels of management including the CIO and the head of whatever business unit that uses the application, and then has to go through the purchasing department. Putting the same app "in the cloud" means charging a monthly fee on the company credit card, approving that takes one boss about 30 seconds.
(Score: 2) by VLM on Tuesday November 25 2014, @05:15PM
Hmm yes but that doesn't protect middle management jobs, so if you self host your own server, you can get NAS, vmware, firewall, security, OS, and apps support all involved to cause more delay than just sending a P.O. to Dell.
Just saying "takes one boss about 30 seconds" is bound to be "optimized" to maximize hiring sooner or later.
(Score: 2) by sjames on Wednesday November 26 2014, @03:19PM
The funny part about cloud is it's becoming popular and widely available right at the time it's least needed.
I would have loved the cloud at one time (if it was properly maintained...), but these days IPMI typically actually works, expensive fibre channel is replaced by inexpensive AOE, common server class hardware (and a lot of desktop hardware) easily supports VMs with a fairly small performance penalty. It's a long way from the days when a kernel update meant you give the reboot command and then have a few minutes of white knuckles waiting for a ping to return. With all of that, I think of the cloud as primarily useful when temporary capacity is needed.
(Score: 3, Interesting) by tibman on Tuesday November 25 2014, @03:28PM
My company has nearly zero local infrastructure. No servers or cable cabinets. There is just a cable internet connection and wifi. All employees (around 30) have laptops so they can work from wherever they want. Phones are voip and easily routed elsewhere. The whole building could burn down and we'd still have a company.
Is this better? I doubt it. It seems more like extreme outsourcing. Only the physical aspects are outsourced though. Developers still have to admin the servers. So i'd agree that it may have revolutionized the way we work but i don't think it changed how products are brought to market. The linked article pushes some management techniques that lead to faster product development. Iteratively developing products instead of large development projects is a big reason. If a team doesn't have to ask permission to implement a feature then of course that feature is going to come out fast. As long as the team has a good feedback mechanism then i see no reason why they need a traditional hierarchy sitting on them.
SN won't survive on lurkers alone. Write comments.
(Score: 3, Interesting) by cmn32480 on Tuesday November 25 2014, @06:23PM
Interesting. We are the exact opposite. We have zero cloud infrastructure, with the exception of our upstream spam filter and our phones.
Management here is not jumping on the cloud bandwagon (nor am I), and we host it all internally. It is their preference to have control of their data, rather than outsource it to a hosting provider.
If our building burns, we fire up the DR site at another office, change some DNS records, and start rolling again.
"It's a dog eat dog world, and I'm wearing Milkbone underwear" - Norm Peterson
(Score: 2) by tibman on Tuesday November 25 2014, @07:50PM
The cool part is your infrastructure probably looks like a cloud to your other offices. They have to remote into the machines to do anything. You might be doing the same to spare yourself a walk down to the physical keyboard and monitor (i would, lol). DR is fine as long as it is a full machine backup. Doing just data and source code really blows. Having to rebuild a full stack to the exact same working environment as before is very tedious. SSH/SSL certs, php.ini, web.config, pre-reqs, specific library versions, yuck.
SN won't survive on lurkers alone. Write comments.
(Score: 2) by frojack on Tuesday November 25 2014, @10:30PM
The agencies I do work for (state government) have an "inhouse cloud". Its all virtual hardware but its all on in-house servers.
It has made some small changes to the way our software has to manage resources, because regardless of how transparent they claim it is, stuff changes behind your back, and the workload you were promised 4 virtual processors to deal with one day drops to two, or your drive space suddenly can't handle those surge loads, and user start wondering why stuff is slow.
Even if they make no changes to your VM's provisioning, they will move it to a blade that is overloaded to make room for some more clout-full project.
No, you are mistaken. I've always had this sig.
(Score: 1) by lizardloop on Tuesday November 25 2014, @05:06PM
I run a small software company and everything we do is building applications to run on "cloud" servers and then administering said servers. I like it because it keeps startup costs. If we aren't sure how busy a particular application will be we can simply launch it on the cheapest VM possible and see if it gets popular launch it on a bigger one. The up front cost is minimal.
So far reliability hasn't been too bad either. We put some work in to making sure a single VM going down doesn't impact us too much and our uptime has been pretty good.
(Score: 2) by VLM on Tuesday November 25 2014, @07:10PM
We put some work in to making sure a single VM going down
How about a single VM provider not just a single VM? Either ditching you as a customer or suddenly closing.
(Score: 2) by tibman on Tuesday November 25 2014, @07:33PM
Just as bad as relying on one ISP for connectivity.
SN won't survive on lurkers alone. Write comments.
(Score: 1) by GWRedDragon on Tuesday November 25 2014, @09:57PM
At least if you have one ISP and they drop you, everything doesn't vanish in an instant. Recovery would be easier.
Cloud services tend to be turned off in the dark of night with little to no warning. It's the cloud way.
[Insert witty message here]
(Score: 1) by lizardloop on Wednesday November 26 2014, @12:37PM
Good question.
For the first couple of years while we had few customers we used a single provider and a single data centre. In the last six months we have started using a second entirely separate provider with a data centre in a very different geographic location.
Yes, if one of our providers suddenly turned off all our VMs it would cause some disruption but we have backups of everything important and could redeploy on the remaining provider relatively quickly.
(Score: 1, Insightful) by Anonymous Coward on Wednesday November 26 2014, @12:42AM
I am a UNIX systems administrator.
I've been doing this for thirty years.
Over the past ten years this cloud thing has knocked the feet out from under everyone over 30.
If you can't prove you can do object-oriented development in Python with a Git repository full of source code, use Puppet and/or Chef, and program in Ruby, you may as well go die.
By confusing skill with a particular tool, with the wisdom that comes from being a tool user for several decades, Silicon Valley has thrown out the baby along with the bath water.
Google led the way by sneering at anyone over 30 and, instead of employing these people and keeping their skills current, they led the way in rejecting anyone who had not graduated from college in the past five or ten years.
There's simply no possible way that a person could simultaneously earn enough to remain fed and housed in the Bay Area AND simultaneously learn two new programming languages ... two new software tools intended to manage large populations of servers ... three or four different contending virtualization mechanisms ... etc.
It's clear. Silicon Valley hates people who have children or significant others. Like a jealous hag, Silicon Valley will not rest until it has taken my entire life - every minute of it - and, somehow, monetarized it, to my disadvantage.
The 'cloud' epitomizes everything I have come to hate about Silicon Valley. In two small syllables - 'the' and 'cloud' - managers have managed to reduce the enormous complexity of an infinitely scaleable, massively robust infrastructure into something a five-year-old can draw.
Then, thinking that, i they - like five-year-olds with a whiteboard - can draw it ... they must be able to understand it ... they then make decisions, based upon this child-like understanding of the Internet ... and then, like children, when it doesn't work, they go into a rage, and demand that someone be punished.
I'm tired of being punished.
They can go build and manage their own fucking infrastructures.
Call me when you have an IT budget that includes continuing education for all the toys, and a retention policy that doesn't treat employees like recyclable cans.
(Score: 0) by Anonymous Coward on Wednesday November 26 2014, @02:58PM
Can I get an "A-Men Brother!"?
(Score: 2) by Geotti on Wednesday November 26 2014, @03:53AM
What is this butt [mozilla.org] people keep talking about in the meeting rooms?
If it's like that trend in Hollywood movies to include more and more toilet jokes, I don't find it funny at all when management suddenly wants to save all their data in my butt.
Like the CIO's not been raping me for years anyway... But enough is enough, he can go and find someone else to administer his butt!
Jokes aside, the only difference to grid computing is the transparent "elasticity" of resources, so it'd be great if someone would translate the intro chapter of Tannenbaum's Distributed Systems to marketing-speak and make it clear to those PHBs that unless you absolutely require a certain degree of transparency (~ies), you're probably better off getting a regular VPS or colo from a reliable provider (or an old server on ebay to put it on premise).
It's an old debate, of course, but really there are few use-cases for SMEs that warrant risking the loss of critical infrastructure or personal data to a glitch in the system, a geopolitical conflict, or whatever.
I mean, how stupid is it to choose a web-based^H^H^H^H^Hscale point of sale of dubious reliability instead of a proven, rock-solid system that's 100% under your control?
Why don't these people understand that paying 30 bucks a month can end up costing way more than getting an admin, a support contract (or someone to set the system up properly and be on call when SHTF). So many SME managers are totally brainwashed by this marketing bs, it ain't funny anymore.
Why do so many people not understand that smart edges are way better for most of them than powerful, centralized cores?
Anyway, to come back to the question, things haven't changed much for me. I just have to work with yet another abstraction layer where things can go awry. Also, instead of the salesmen from HP/Dell/IBM I have to work with the salesmen from Amazon/MS/HP (at least one stayed the same).
There are some cool tools like openQRM, though that can help you manage instances and move them almost freely between public and private resources (i.e. hybrid cloud), so all in all, it's just grid-computing evolved for me.
(Score: 2) by aclarke on Wednesday November 26 2014, @02:06PM
We've moved entirely to the cloud and it's great. When I came on board to my current company three years ago, all our physical servers were hosted in a data centre in another country. At some point we reached the point where nobody who still worked in that city had the technical capability to actually manage those servers. Even getting a new drive installed was a hassle and an expense. Our servers were running old and unsupported versions of Linux, and the system was so fragile that we couldn't safely upgrade them and we didn't have the hardware to move to while the old servers were down. It was a miracle everything was still running and it was just a matter of time before everything came catastrophically crashing down.
We backed up everything and migrated to AWS. We save money every month, and we now have a scalable and more fault-tolerant systems architecture. Not everything has been roses and caviar in the transition, but because I've used AWS extensively in the past I knew what I was getting into. I'm in my 40s with a young family and significant responsibilities outside work, yet I still manage to make time to keep up with new ideas as much as possible. I am fortunate to have a very supportive work environment to help me with this. I get a little tired of hearing almost everyone on this site rain on The Cloud every time the idea comes up. It's not the right choice for every organisation, but for many products it's the clear and obviously correct choice.