Are Network and System jobs going to disappear?

nact

I heard that because of Cloud and DevOp, the network and system jobs are going to disappear. How long would it take for it to disappear?

What do you have to do to get a job as in the Cloud field?

N7Valiant

Sounds like a hype train.

I live in Hawaii with island internet. "The cloud" means a datacenter down the block. We recently had issues with them with one client having messed up index issues caused by a change of hardware that their terminal server runs on forcing us to disable indexing entirely. Another client had them hosting their servers but somehow had the server and the backups become corrupted. A 3rd client is getting next to zero service from them as they are managing their e-mails.

Needless to say, all 3 of them are reconsidering their business arrangement. I also find it hard to imagine having say, a datacenter host things on "the cloud" server wise and not running into constant latency issues.

As for getting a job in the cloud field I might think living near a datacenter for a cloud provider would be a pretty big one.

LordQarlyn

I have to agree with N7Valiant. The cloud is here and here to stay. That said, there are issues and concerns for a business to cede total loss of control to an outside party that has their own vested interests. Once the "ooh" factor of the cloud fades, businesses will realize putting everything in the hands of a 3rd party may not be the best idea, and will bring back some IT functionality to themselves. The cloud will always be a big part of IT in the future, and there will be inherent advantages to using the cloud, but I don't think it will totally disrupt the IT industry as we know it.

Finally, even at cloud datacenters, there are servers who need system admins, engineers, and architects, and network infrastructure, needing network admins, engineers, and architects. Those are not going to go away. If you want a job at a cloud provider, work on datacenter related certs.

Info_Sec_Wannabe

Agree with both.

To add on LardQarlyn's point, giving total control to an outside party would be a big no-no for those industries that are heavily regulated and publicly traded...

LordQarlyn

Info_Sec_Wannabe wrote: »

Agree with both.

To add on LardQarlyn's point, giving total control to an outside party would be a big no-no for those industries that are heavily regulated and publicly traded...

Hahaha thanks for reminding me I need to get back into my gym routine

MeanDrunkR2D2

They aren't going anywhere at all. For many Sys/network admins racking and hardware parts will become a thing of the past for most places outside of cloud datacenters or companies with highly risky stuff that may have to stay on site. Now, what will a Sys/Network admin do after that? For the most part the same stuff as today with applying patches, adjusting performance on those systems, working on bigger projects and most likely needing to do more scripting or using containers to automate many day to day tasks that don't need human interaction as a general rule. Things will always break and need to be fixed. The titles won't go away, but what a Sys Admin did in the Server 2003 days to what they'll be doing in 5 years will be light year apart in the day to day stuff.

EANx

The connections to the cloud aren't going to disappear. Whether the data is on-prem or off, users still have to get to it which means networking. Users still need provisioning and access control and sub-groups within the organization need someone managing those shares which means system skills.

paul78

Info_Sec_Wannabe wrote: »

To add on LardQarlyn's point, giving total control to an outside party would be a big no-no for those industries that are heavily regulated and publicly traded...

I'm not sure about other countries but that's not true in the US. Many public companies and companies with regulated data are either moving to cloud based solutions or utilize service providers which are cloud based. For many companies, running a data center is generally not a core business competency. And for smaller service providers including those that process regulated data, using cloud services are common place if those companies are to remain competitive.

It's generally a fallacy to believe that in-house solutions are more secure or better managed than a cloud solution. Cloud solutions such as AWS offer far more resiliency, elasticity, and security controls than most companies are able to accomplish on their own.

Also - the notion of "Cloud" as a third-party service provider isn't particularly novel. Companies have been out-sourcing infrastructure for as long as there have been computers.

@OP - as others have mentioned - network and systems are not going to disappear. Especially not server admin roles - cloud iaas still needs people that actually have to build, deploy, and manage those servers.

What will change is the skills needed to be a competent tech admin - Cloud/DevOps/(insert your favourite title) are simply ways to describe roles where basic sysadmin and netadmin skills are used in different ways. One thing that has changed is that sys/net admins are expected to scale well because there are better tools available for automation.

That all said - technologies which provide FaaS will increase infrastructure efficiency and many SaaS providers will have less of a need for dedicated sys/net admins.

BTW - The number of jobs may not necessarily go down, what's probably happening is that there are more people that enter the field because the level of in-depth technical knowledge has gone down.

Node Man

The cloud is just someone elses computer. Still need to connect to it. Still need to keep it safe. Still need to back it up. It still needs to run services.

ccie14023

This question comes up like every two weeks. Allow me to self-promote.

On Being a Dinosaur - SubnetZero

...and...

The value of a CCIE - SubnetZero

NetworkNewb

We'll be on a Universal Basic Income and almost all jobs will disappear as our AI robot overlords take over all the jobs anyways...

But really like Paul said, jobs adapt. These new titles are just jobs transitioning and looking for new skills with new technologies. SysAdmins and Network Admins aren't going anywhere for awhile Their exact skills and what they do will just be ever changing. Just don't become complacent with one technology... Try and keep your skills up to date and you'll do fine in either field.

PCTechLinc

Check out the PS4 game "Detroit Become Human". It's set in Detroit (obviously) 20 years in the future, where almost all jobs are done by Androids and the unemployment rate is like 40%. Pretty interesting story.

N7Valiant

PCTechLinc wrote: »

Check out the PS4 game "Detroit Become Human". It's set in Detroit (obviously) 20 years in the future, where almost all jobs are done by Androids and the unemployment rate is like 40%. Pretty interesting story.

Also kind of assumes you don't need humans to develop and troubleshoot those androids. Because if computers never broke, you wouldn't need a help desk.

LordQarlyn

I'm all on the cloud bandwagon, I use cloud services extensively. And in very technical sense, the early computer systems in the 50s and early 60s could be described as a form of cloud computing. Certainly the mainframe was like the cloud provider and end users accessed it through dumb terminals, paying for processor time and other systems resources on a per usage basis. The modern cloud computing solutions offer a lot of advantages, very scalable, very flexible. But moving to the cloud means making trade offs just as with anything else in life.

Of course it is not an either/or. I think many bigger organizations will have a combination of public cloud and private cloud coexisting along side their own existing enterprise architecture. So, even as Amazon managed to convince the DoD it's just fine to store classified information up to Secret on AWS a system that's used by and serviced by many uncleared people some of whom are from countries who adversaries with the US, the DoD is still maintaining, expanding, and upgrading their existing enterprise network.

paul78

LordQarlyn wrote: »

So, even as Amazon managed to convince the DoD it's just fine to store classified information up to Secret on AWS a system that's used by and serviced by many uncleared people some of whom are from countries who adversaries with the US, the DoD is still maintaining, expanding, and upgrading their existing enterprise network.

Yeah - great points - AWS GovCloud being available for DoD is a good reminder about how far Cloud adoption has come - https://aws.amazon.com/compliance/dod/ and https://aws.amazon.com/govcloud-us/dodsrg/

networker050184

Learn and evolve because technology moves fast. You don't want to be one of the ones left behind.

dehgrah

Systems Administrator jobs are not going any where they are in fact transitioning to DevOps but keep in mind it's essentially the same job role except with additional responsibilites. DevOps must have the knowledge and skill of a Systems Administrator but also have a baseline knowledge coding, scripting or both. DevOps also ties the relationship between IT Operations and (Software)Development.

Network Administrator jobs are here to stay as well but again with SDN taking over you must pick up additional skill sets such as coding or scripting using Python. A python programmer can not just become Network Administrator because hey know how to code, they must also learn the OSI, Subnet, and Protocols of how Networking operates.

Don't think too much about this, we years and years away from AI taking over and even then who will maintin advance OS's and AI's? See where I'm getting at? It's always a need for someone to keep "Man Made" technology up and running.

LordQarlyn

paul78 wrote: »

Yeah - great points - AWS GovCloud being available for DoD is a good reminder about how far Cloud adoption has come - https://aws.amazon.com/compliance/dod/ and https://aws.amazon.com/govcloud-us/dodsrg/

Yeah, exactly! Storing unclassified but sensitive and classified information on a public cloud. What could possibly go wrong with that?

paul78

dehgrah wrote: »

we years and years away from AI taking over and even then who will maintin advance OS's and AI's? See where I'm getting at? It's always a need for someone to keep "Man Made" technology up and running.

That's exactly what the AI's and robots who are building the Cloud want us to think.

dehgrah

Man made objects always fail and break. Advance advance software will always need constant updating including the hardware it's on.

nact

Systems Administrator jobs are not going any where they are in fact transitioning to DevOps but keep in mind it's essentially the same job role except with additional responsibilites. DevOps must have the knowledge and skill of a Systems Administrator but also have a baseline knowledge coding, scripting or both. DevOps also ties the relationship between IT Operations and (Software)Development.

I see.

Network Administrator jobs are here to stay as well but again with SDN taking over you must pick up additional skill sets such as coding or scripting using Python. A python programmer can not just become Network Administrator because hey know how to code, they must also learn the OSI, Subnet, and Protocols of how Networking operates.

Can you learn node.js instead of Python?

Pmorgan2

Cloud Services and programming are tools system administrators should add to their toolbox, but I'm not worried about the rest of my tools rusting from lack of use.

DevOps positions will eat into some system/network administrator positions. However not all organizations can benefit from DevOps no matter how much their new CTO wants to put a cloud migration on their resume. Not all organizations are development focused, appropriately flexible, or comfortable with the limitations of cloud platforms.

For example, my organization has developers but they just another support group for our organizations main goals. One of us system administrators ought to migrate some of our infrastructure to the cloud, but that would still leave a lot of infrastructure for the other system administrators to maintain / improve.

Anecdotally, the biggest thing holding us back is internet speeds. To consider using a cloud service, we'd need a few 40 Gbps connections and several 8 Gbps in order to keep up with our data processing. Then we'd need to get a good price on >3 Petabytes of storage including disaster recovery.

We pay out our as$ for less than 1% of that speed. 3 Petabytes on Amazon is 6x more expensive than our on-premise SAN including maintenance, labor, and utilities. I don't see us moving to the cloud for awhile.

paul78

nact wrote: »

Can you learn node.js instead of Python?

Node.js is not typically used for administration. It's a good for developing web apps if that's what you are interested in doing. Other languages you can also explore which are common for administration include ruby, perl, powershell (windows), and .

Panther

paul78 wrote: »

BTW - The number of jobs may not necessarily go down, what's probably happening is that there are more people that enter the field because the level of in-depth technical knowledge has gone down.

I'm not in DevOps (thinking about it), but hearing what their backgrounds should be (sys admin, networking, scripting), I'm thinking the veterans in the field (former sys admins, network admins) would fair better, versus the newcomers.

I almost think newcomers will have harder time because they don't have the in-depth knowledge, accumulated over the years.

Then again, maybe that's what employers are looking for now, as DevOps you know a little of everything from sys adm, networking, scripting, etc.

LeBroke

The takeaway from this thread is "learn to code lul"

All jokes aside, the cloud is an absolute killer in some industries, borderline useless in others, and somewhere inbetween for the vast majority of businesses.

I feel it's important to separate cloud into two broad categories - SaaS (Software-as-a-Service) like Google Docs, Paypal, Zenefits, or ADP, and IaaS (Infrastructure-as-a-Service) like AWS or Azure. There's also PaaS but I consider it a subset of IaaS which abstracts away the virtualization/OS layer.

Warning: long essay incoming.

tl;dr: cloud is great if you're a small business or running a pretty standard webapp.

SaaS

SaaS services are revolutionary for small businesses. Why bother running an email server if your company has all of 10 people in it and no dedicated IT guy? Google for Work or Office 365 can provide you with a great set of tools that requires little management, supports your own domain/branding, and is multitudes more stable than anything a shitty domain hosting provider can do.

This does decrease the need for conventional helpdesk/sysadmin/MSP duties as companies don't really have to think much about IT until they grow to a certain size. As loath people here would be to admit, but you can run a small business perfectly well off Gmail + Google Docs, a couple of PCs, and a SOHO router, especially if technology isn't their area of focus. Just don't save anything locally and you're golden.

There is a lot more chance your 5 year old Windows 2008 domain/print controller that doubles as a fileserver will catch on fire (or an overzealous cleaning guy will accidentally drop a heavy box on it) than Google losing your data.

Same goes for vast majority of business needs that aren't relevant to their primary business goal. A construction company doesn't want to run a large accounting or technology department, for example. It's the same as a tech company bringing in external contractors to remodel their office instead of finding an employee that does woodworking on the weekend to do it.

IaaS

IaaS services, on the other hand, are revolutionary for, well, SaaS products and webspace in general. Until you get to the scale of Facebooks and Spotifys of this world, IaaS like AWS or Azure are going to be both cheaper and easier. Cheaper because despite the sticker price tag, there's tons of ways to reduce cost. Your demand is extremely spiky based on consumer habits (i.e. everyone watches Netflix in the evening, but few people do at 7 AM)? Great, utilize autoscaling. Get big enough and you can write scripts, crunch data, or even utilize ML to predict demand and scale out in advance instead of in response to demand, and then scale back down once peak hours pass.

Need storage? Sure, you can use EFS or store stuff on EBS volumes, but that's going to be pretty expensive, and either slow, or with a single point of failure, respectively. If only there was some kind of cloud storage that could store objects on the cheap... Sure, it's going to be fairly slow if you want to retrieve all your data at once compared to a SAN, but I hazard to guess most application workloads don't actually need 100% of their data available all at once (if they do, it's probably a database to begin with).

Storage Example

Here's a quick and dirty calculation for you. 2.5 years and 2 jobs ago, we were looking at buying a new SAN for our SaaS product. NetApp quoted us about 42k USD for a 30 TB SAN (with a 3 year support contract). If our application was able to utilize S3, it would have cost us $0.03/GB * 30000 * 36 months = $32400, and would actually get cheaper as time went on (S3 pricing is $0.023/GB now). While slower, it would also be an order of magnitude more resilient since all data would be duplicated at least once, and would make any data migrations easier since S3 is accessible from anywhere you have an internet connection, whether or not you're using any other AWS services.

We only had 10 TB of data at the time, so we would be paying for 20 TB of space that wouldn't be utilized for at least 1-2 more years at our rate of growth. The SAN would also have a hard limit of 30 TB, so in 2-3 years we would need to add a second SAN and then figure out some way of clustering them together.

Reliability

For reliability it's at best a wash, but generally favours T1 cloud providers (AWS/Azure/GCP). You're still either running your own data centre, using colocation, or renting out servers. If you're running your own datacentre, you now have a lot of overhead, and likely a lot less money to spend on resiliency, backups, a secondary site, and the ilk than Amazon does. I honestly trust Amazon a lot more than I trust a barebones crew to keep the lights on because they have both the economy of scale and a business need to make everything as resilient as possible. A company running 3 racks in their basement just won't have the money or expertise to take care of every possible contingency to keep their systems running, whether it's a fire, tornado, or some guy who took a backhoe to your internet connection.

Hell, even mid-level providers have trouble literally keeping the lights on. I've been through I think... 4 critical outages in my fairly short career that lasted a better part of a day. One pretty funny incident - "Sorry for the outage, our primary firewall just went down. We're trying to failover to the standby. Ooops looks like standby firewall isn't configured. We're configuring it now. Double oops, looks like the standby firewall is actually broken too. Hold on guys, Cisco is Fedexing us a new one. We'll let you know in 6 hours." This was a dedicated service provider that was charging us $10k CAD/month for 1 TB of shared SAN storage and like 3 low-end boxes (4 cores/16 GB RAM) in 2015. Pretty sure our CFO was taking kickbacks since he harsh blocked any attempts to migrate to much better and cheaper services, but that's beside the point.

If you're using colocation, it's one of two options. Either you're using a fairly cheap provider, in which case anyone who works there (or in especially bad cases, just walks in through the door) can just stick a USB or attach a KVM into your super critical server hosting HIPAA data or running credit card transactions and go to town. Not to mention there's a 50% chance their generator won't even work. And if you're using a high-end provider like Switch, you're looking at $3,000+ per month per rack. You're still paying for your own hardware.

In 3 years I've been using AWS professionally, I've only seen two outages I could classify as major. First one, S3 was down for a few hours... After which they did a detailed write up of how they screwed up and all the controls they put in place to make sure it doesn't happen again. And second one, just a few months ago, EC2 instances in a single AZ went down and stopped being accessible for about 2 hours. Parts of our application that were multi-AZ weren't affected, and parts that we haven't migrated to utilize multiple AZs yet were impacted in one of our accounts.

Won't get into compliance anymore since this is already probably more than anyone wanted to read.

Disadvantages

It's expensive for static workloads. Requires pretty high engineering expertise (need to code and have good knowledge of architecture even for pretty low level jobs). Requires your application to be architected in a way that utilizes cloud-specific features like S3. Doesn't do much if all you're running is internal services like AD (probably less reliable this way considering your internet can die at any time).

sonyvaio18

Thanks for the awesome tips everyone!

blargoe

Sys-admining and and network-admining the way it has traditionally been done with the traditional toolset and mindset is going away. The jobs themselves are not.

EANx

Every 5-7 years, something comes along that makes everyone believe "this is a game changer, get on board, your jobs will be going away" and they're sometimes right about the first, but not the second. Virtualization, eCommerce, the Internet, client-server, etc. As long as your users are in one place and the resources they need are in another, there will be a need for network and system admins.

LordQarlyn

EANx wrote: »

Every 5-7 years, something comes along that makes everyone believe "this is a game changer, get on board, your jobs will be going away" and they're sometimes right about the first, but not the second. Virtualization, eCommerce, the Internet, client-server, etc. As long as your users are in one place and the resources they need are in another, there will be a need for network and system admins.

Heck as long as there are networks and networks connecting networks, net admin jobs are not going anywhere regardless if everything is cloud, on site, or some future not yet thought of arrangement.