Another One Bites the Dust: Cisco Discontinues Its $1B Cloud Initiative as AWS, Azure and Others Expand (geekwire.com)

Cisco will abandon its InterCloud cloud-computing offering on March 31 and will move any InterCloud workloads to other, unnamed cloud providers, including "in some cases, public cloud." From a report on GeekWire: Cisco's pull-back from the cloud scene marks the latest example of smaller participants -- many of them hardware-makers -- bailing in the face of huge growth by Amazon Web Services and Microsoft Azure, and to some extent by Google Cloud, IBM and other, smaller public-cloud services. Hewlett-Packard in 2015 abandoned its efforts to be a public-cloud company. Then, Hewlett-Packard Enterprises essentially shut down its much-ballyhooed Helion cloud offering earlier this year. VMware still offers its vCloud Air hybrid-cloud service, though it has agreed to partner with AWS, which it once viewed as its arch-rival for cloud workloads. "We do not expect any material customer issues as a result of this transition," Cisco said in response to a request for comment. "For the last several months, we have been evolving our cloud strategy and our service provider partners are aware of this."

Comments

The bubble is popping

[guruevi]

It seems companies are finally realizing they've been over-saturating the market with cheap VPS and people are finally starting to realize the security and other implications of shared hosting at a handful of providers.

I don't know if Dyn's outage a few weeks ago finally got the managers to listen and start diversifying their systems again.

Re:The bubble is popping

[Anonymous Coward]

Cloud is bullshit.
Regular hosting is more or less intuitive.
And then there is iCloud.

Re:The bubble is popping

[Anonymous Coward]

I work for one of the major cloud providers listed above. In addition to market saturation, what is also killing them off is the massive capital investment required. The place I used to work for thought hey we will just buy a bunch of Cisco UCS chassis and make a "cloud" environment out of it. It sucked. To get into the major leagues, you have to basically build your own custom hardware... specialized stuff, high performance, low latency, very energy efficient. The cost of the servers is enormous, plus the internal networking, the site to site connectivity costs, the data centers themselves, and electricity. Then... dear God, you want redundancy? The costs just went through the roof. We deploy servers in lots of 5000. It's not trivial.

Unless you can secure a customer base to offset the rapidly increasing costs of deploying a cloud infrastructure, it goes into the red quickly. Not only do you have to win the customer, you have to be price competitive. I don't think many companies understood just how much money they were going to have to pour into cloud infrastructure.

Re:The bubble is popping

[Anonymous Coward]

Okay grandpa. AWS is amazing.

Yes, there are some good reasons to not use it, but overall it's lightyears above managing your own servers or "regular hosting", which is really only viable if you're only working with a few sites, and not completely different environments for every subdomain.

Re:The bubble is popping

[known_coward_69]

i'm sure building the software around it is also a big deal

MS launched it's first cloud service around 2002 or 2003. i was one of the original users when xbox live first launched. they were also learning from hotmail at the time and they had a bunch of other projects at the time that weren't public. i even tested a MS version of dropbox, before there was a dropbox. MS killed it and from what i heard these guys then started dropbox

AWS launched something like a decade ago after Amazon built A9 and some other services before that

you can't just write some software in a year that other companies had been coding for a decade or longer

Re:The bubble is popping

[Anonymous Coward]

If you think cloud solutions like Azure and AWS are "cheap VPS" you really don't understand modern IT.

Re: The bubble is popping

[Anonymous Coward]

Ok, whippersnapper (different 'grandpa' here). I'm not sure if 'amazing' to you means 'has useful functionality' or 'does things I'm too clueless to figure out', but let's go with the first one just for argument's sake.

The things that make AWS amazing are management tools and applications. It's nothing that can't and won't filter down to ownable equipment, if not by them than by a competitor.

See, there's this thing called economics. What you've got right now is Amazon and Microsoft basically becoming a cloud duopoly. When they finish, they'll become a rent seeking duopoly and start raising prices. A lot. Microsoft already does this when you renew Office365 subscriptions. People aren't going to like or put up with this permanently and, unlike with cable and ISPs, there's nothing stopping them from making choices.

Cloud is already not cheaper at any kind of permanent scale, and by that I mean I always need X resources all the time. Sure it helps in those very few and highly marketed edge cases where you have 10 servers and need 100 for a week or so, but most people's businesses don't work like that. It's also good for very small businesses that have nothing, at least as a starter set.

So on prem stuff will become more attractive again as it becomes more capable and easier to manage, or actual cloud competition will develop, or likely both. What will NOT happen is everybody uses Amazon and Azure for everything like the cloud snake oil peddlers keep predicting. We'll settle into a world where people use what's logical for them to use.

Re:The bubble is popping

[btroy]

No. Cloud is expanding. The flexibility and cost savings for test environments and even prod environments on non-sensitive information is just too enticing. I think the person's response of the costs of implementation are spot on. Plus, Cisco (router/network company), not many C-levels are going to take their cloud offering seriously.

MS - made it easy by integrating with their products
AWS - been there from the early days and the same, a lot of tools have ties to AWS built into their product
IBM - They'll succeed simply because of their history and name (not so much reputation any longer).
Google - same thing, tools already have integration and they'll keep trying harder. (Consumers will be their client)

Re: The bubble is popping

[skids]

The other thing that will kill cloud is arbitrary service alterations when the duopoly gets cocky, and the fact that smaller providers can just pack up and leave.

Basically if you use too many privately owned, nonstandard APIs/infrastructure, you are in for hurt when those stop being offered on a whim. It will probably take a few decades before this lesson filters up to PHBs, but someday there will come a buzzword down the tubes that means "stop using any old thing a company offers and stick to standards."

Re:The bubble is popping

[skids]

on non-sensitive information

What's that? Never heard of it.

Re:The bubble is popping

[Anonymous Coward]

Plus, Cisco (router/network company), not many C-levels are going to take their cloud offering seriously.

Nobody ever got fired for recommending AWS.

Re:The bubble is popping

[Anonymous Coward]

How good do I understand modern IT if I think these "cloud solutions" are the term marketing is using this month for good old data centers?

Re:The bubble is popping

[guruevi]

Okay, they are expensive VPS with an API to generate more VPS from a barebones image. Once you get down to the metal, there is actually nothing new or complex to the whole cloud thing, yes, the programs are more complex to bootstrap a virtual datacenter instead of a single virtual server but there is no 'special sauce' going on that turns it into a self-aware entity that knows how to obtain more resources.

We had a similar setup way back when I first entered the industry at a dotcom hosting company, I was one of a total of 4 engineers and 4 programmers (about 12 staff total) operating 2 fully owned data centers and 3 shared data centers, each had several thousand nodes, being backed up, deployed on demand etc. We rented out dedicated machines, virtual machines (containers) and shared hosting.

Re:The bubble is popping

[Anonymous Coward]

You "VPS with an API" is about 1/100th of what Azure and AWS are offering. You're comparing apples to oranges. If all a company is doing is moving their servers to a hosted server somewhere, they are missing out on a whole lot that Azure and AWS can provide.

I too worked for a dotcom hosting company in the late 90s. What they were doing then compared to Azure or AWS do today is night and day. Do both solutions sit on bare metal somewhere in a data center? Sure, but that is where the comparison stops.

Re:The bubble is popping

[haruchai]

"We had a similar setup way back when I first entered the industry at a dotcom hosting company"

What was the company's name?

Re:The bubble is popping

[Cramer]

you can't just write some software in a year that other companies had been coding for a decade or longer

You can if you believe the BS from Rackspace and Facebook... OpenStack! (hint: you'll die of old age or commit suicide long before you get anything remotely usable based on openstack) You can quickly setup a "cloud" using VMware's collection of purchases, but you'll go insane trying to make sense of it all, and end up bankrupt.

Much Ballyhooed?

[ickleberry]

Does this refer to Ballyhooly?

now if they could just axe their servers.

[nimbius]

Its a little off topic but its got to be said. Servers that do every single conceivable out of band management thing in the world. I blame dell for this, but Cisco has taken it to a whole new level. Weve had IPMI and SOL over BMC for sixteen glorious years, and thats worked fine to ditch the console servers and the overpriced intelligent power wips in the datacenter. But cisco's UCM blade platform is a fever dream of browser based garbage designed to configure everything from the servers network IP and route, to inventory and asset management. im sure this is great if your datacenter is a single vendor, but in real datacenters there are about two dozen of these kinds of products in constant battle with eachother. Each has their own plugin, interface, configuration workflow and god help us configuration language. Ciscos UCM is a committee based piece of garbage.

in the real world this is nothing more than firmware-based bloatware. it frankly drives me away from buying from these vendors that cant just deliver the hardware as they always have without engaging in some value-added fart huffing contest to see who can create the biggest branded cockup. Vendors should take note that all this garbage just gets shut off and ignored in favour of IPMI and Salt/Chef/Ansible/a sensible configuration management solution that isnt tied like some rented mule to a multinational corporations committee based meth-addled future predicting marketing department.

Re:now if they could just axe their servers.

[thomn8r]

UCM blade platform is a fever dream of browser based garbage designed to configure everything

The desired goal is vendor lock-in. HP(e)'s OneView makes UCM look downright sane.

Re:now if they could just axe their servers.

[Anonymous Coward]

Real Datacenters uses http://opendcim.org/ or something like it. (shill warning: I know the developers) that is OSS, headed up by 2 former Vanderbilt University Datacenter ops employees who have made this tool to not only help themselves but others. It is even used at CERN to keep track of the servers collecting all the HLC data.

Re:now if they could just axe their servers.

[Anonymous Coward]

Agreed, openDCIM (Data Center Infrastructure Management(because parent did say what it actually was)) is what we use at my work too.

Re:now if they could just axe their servers.

[Anonymous Coward]

I have to agree with you that their server offering is more convulated than it should be (and than what their sales pitch is telling you). We're Cisco Partners, so we had their FlexConnect solution at a fair price. As I switched from systems to networks some time ago, I only have older generations of servers from HP and IBM to compare it to but those were much easier to set up and manage than Cisco's and the vocabulary was much more straightforward.

Worse is how buggy the packages doing all the magic are. Firmware and drivers that don't update or install, drives that aren't detected, an HTML5 interface that doesn't always updates the information being presented or asked for... Good thing Ansible is hiding behind their FlexConnect Installer so I could check it out and complete the steps manually or we would have spent more weeks with a TAC engineer who might be a CCIE but clearly was overwhelmed when it came to servers.

Of course, as a Cisco Partner, I'm posting this anonymous.

Re:now if they could just axe their servers.

[skids]

value-added fart huffing contest

I am sooooo stealing that.

Heavy clouds

[Anonymous Coward]

This is precisely why I don't rely on the cloud exclusively. It could be any company on any day, and your data is just history. Poof! We simply can't rely on third parties for everything, it isn't realistic and it isn't smart. We need to be the arbiters of our own lives and affairs, not Facebook, Google, Uber, Amazon, Evernote, etc., ad infinitum.

Re:Heavy clouds

[luis_a_espinal]

This is precisely why I don't rely on the cloud exclusively. It could be any company on any day, and your data is just history. Poof!

And that's why backups exists (specially incremental backups). Cloud providers (large, reputable ones, that is) do not disappear within 24 hours. They give you plenty of time to offload to another provider or to physical storage.

And with providers like AWS, redundancy pretty much nullifies most forms of data loss. Any data loss that you experience will most likely be a function of your application or your data management policies.

AWS and Azure

[Anonymous Coward]

Azure for .NET environments. AWS for everything else.

There's little point in getting involved with anything else. It might go under, the company might lose interest and stop adding new features, etc.

Pick AWS or Azure. Preferably AWS.

"vCloud Air hybrid-cloud service"

[DarthVain]

Well there's your problem VMware, you really need a catchier name than that! Not sure who come up with the current but it is terrible.

Also "For the last several months, we have been evolving our cloud strategy and our service provider partners are aware of this.", I'm not sure it's "evolving" if your plan is to discontinue it. Extincting might be a better word (if that even is one).

Re:"vCloud Air hybrid-cloud service"

[quetwo]

I guess the positive side of VMWare is that whey they pick a name, they usually only stick with it for about 6 months or so.

Damend if you do, damned if you don't

[Anonymous Coward]

Less about capability, more about lack of standards. IPMI works very well because it was an exceptionally specific standard, that encompassed the requisite functionality without wiggle room.. Just like SNMP mibs developed in the late 90s were nice and specific (and even then, Cisco ignored many of those in favor their proprietary mibs).

In this century, the vendors have taken back control of the newer so-called management 'standards' and make them all terrible. Netconf, CIM, Redfish, all terrible. They all prioritize the ability of the vendors to 'differentiate' to the point of making it useless for developing cross-vendor with a single set of code. So it empowers vendors to convince people they are writing to standards when in fact they are writing locked-in automation, Note that all these standards are comprised *entriely* of vendors, with no customer represenation.

On the flipside, OpenCompute *could* have been something to keep vendors in check. However, in practice only the extremely big companies get their way in designs that don't help the larger industry, and no two customers use the same standards limiting the upside for vendors to comply even if the designs were good.

This would be fine except for the fact that there is pressure to torpedo working standards like IPMI, because 'progress'.

Block all AWS

[Anonymous Coward]

My firewalls block anything incoming from AWS. It seems that 90% of AWS customers are hackers running penetration scripts against various networks.

I wonder why

[slashmydots]

You know why everything Facebook launches is a flop? Because everyone hates them. Cisco has a reputation for costing waaaaaaay the hell too much money for basically everything. So why would anyone let them hold their data hostage for whatever price they demand?

Re:I wonder why

[Anonymous Coward]

So why would anyone let them hold their data hostage for whatever price they demand?

Because they are still cheaper than Oracle???

Re:I wonder why

[Voyager529]

You're absolutely right in that Cisco is not a company with any concept of how to compete on price. I'll do one better though - there's no integration with anything, and thus no reason to use Cisco over AWS/Azure/GCC. Those companies have massive scale, and playing catch-up isn't cheap without a reason to not just use one of them. VMWare can do hybrid cloud better than Google can, so they can successfully charge a bit more to companies who need certain things on-prem while cloudifying others as they decommission old servers. Oracle is run by Satan himself. Data that ends up in an Oracle database doesn't come out barring a miracle, so they can sell to companies who have already sold them their soul pretty well. Speaking of companies run by Satan, Intuit can do their SaaS thing because it's still pretty profitable to have 80% of Main Street businesses dependent on their product, to the point that raising the cost of their Quickbooks Online by $1/year will never be a reason to switch away from them, but is still millions a year in money-for-nothing.

Cisco sells routers and switches and phone systems. There's no springboard, no inertia, no existing customer base to move to the cloud (maybe the phone systems a bit). If they can't leverage their existing customers, then they are left competing with Amazon, the Wal-Mart of the internet, Microsoft, who can still flex muscle with Windows and Exchange server and make those cutovers easier than anyone else, and Google, the company who data mines for a living. Cisco can't compete on price, they can't really compete on features, and they don't have an explicit market for which keeping with their existing vendor makes sense for IaaS.

Agreed & oddly how do I know this?

[Anonymous Coward]

See subject: Building my custom hosts file (blocks more threats vs. firewalls as most = served by hostname, not IP address) using APK Hosts File Engine 9.0++ SR-4 32/64-bit https://www.google.com/search?hl=en&source=hp&biw=&bih=&q=%22APK+Hosts+File+Engine%22+and+%22start64%22&btnG=Google+Search&gbv=1/

I've seen domain/host name data since 1997 for the purposes of protecting users vs. threats online & I recognize when a 'trend' begins from various hosts online being abused due to it.

* Oddly, malwarebytes' personnel (who host & recommend it @ their hpHosts site, a source of custom hosts data) uses AMAZON for their daily dynamic updates (they had to - hpHosts' regular servers couldn't handle the loads of demand for their data - which oddly (not) coincided w/ my program being featured there).

APK

P.S.=> You're right though - MANY 'cloud' system are being rampantly abused for no good... apk