Slashdot Mirror
Verizon Launches Auction To Sell Data Centers (reuters.com)
operator_error writes: Verizon has now chosen to reverse "its strategy to expand in hosting and colocation services after it acquired data center operator Terremark Worldwide Inc in 2011 for $1.4 billion", and has "started a process to sell its data center assets". The so-called 'colocation' portfolio up for sale includes 48 data centers, and generates annual earnings before interest, tax, depreciation and amortization of around $275 million. The enterprise telecommunications industry has had to adapt in recent years to corporate customers seeking more sophisticated and cheaper offerings to manage their data. Verizon joins a host of its rivals in telecommunications who are shedding their data centers.
Most datacenters have a large number of carriers with connectivity into them, regardless of who owns them.
I bid $1
require an NSA risk adverse buyer.
Data centers suffer from very rapid capital depreciation. Kryder's Law means your storage hardware loses half its value every 13 months. It is a very competitive business, and only large scale automated data centers can be competitive. This is the start of the shake out, not the end. You will soon see more companies exit the business.
They bought spectrum which is the property of the people, agreeing to use it fairly. Yet they prohibit devices on their net and require them to go through long "testing" processes that can take up to a year - on devices that have two year shelf lives. They are also against net neutrality.
I figure if it is bad for Verizon, it is good for the public in general.
This is a sign they plan on taking net neutrality seriously... at least for now.
I wouldn't be too surprised if Microsoft bought some, gutted them, and fitted them with new racks of up-to-date gear. Right now MS is desperately in need of physical server space because their Azure stuff is actually (surprisingly) getting a lot of traction.
They literally cannot build datacenters fast enough so they've taken to leasing buildings and then doing the gut/harden/refurbish thing to them, turning them into colos and full-fledged DCs.
An existing DC would be a prime candidate for this as it's already a DC and would only need modern racks and servers. Cooling, power, and physical security are already there and that's what takes the longest to complete. Rack installation, on the other hand, can be done very quickly.
Just cruising through this digital world at 33 1/3 rpm...
Verizon is also shedding land lines...
The problem with "edge" data centers is you need nearly the same operating staff as you do for a large "core" facility, and it is hard to get the same efficiencies. My business model from 2005 is about to take off though, unfortunately without me being able to capitalize on it.
Obviously high- end networking hardware DOES need to be replaced pretty often, perhaps every 5 years or so.
The claim that an enterprise storage system loses half its value in a year because drives get bigger is a bit silly, though. Several years ago I bought storage hardware for our datacenter. It was filled with 400GB drives. The same sleds now hold 4TB drives, connected to the same expanders, the same RAID cards, and the same storage logic. The actual hard drives are less than half of the cost of the total system.
Assuming a five-year lifespan, you'd figure 20% per year on your networking equipment. A bit longer on your racks, even longer for your raised floor etc.
And what is that business model? I did my get it from the context.
Chance favors the prepared mind.
Perfect is the enemy of good.
Obviously, Verizon will simply rent data center resources from the new NSA data center in Utah. It will make sharing our information much simpler.
It must have been something you assimilated. . . .
Containerized non field-serviceable data centers that operate in clusters of 4-6 nodes. At the time it would go to 500kW per node; not sure what is economically viable today though. Swap nodes every 30 months and replace bad parts in a service center.
It was innovative but impractical in 2005... today not so much (on both counts).
Data centers suffer from very rapid capital depreciation. Kryder's Law means your storage hardware loses half its value every 13 months. It is a very competitive business, and only large scale automated data centers can be competitive. This is the start of the shake out, not the end. You will soon see more companies exit the business.
Until someone tells them data centers = cloud infrastructure. Seriously, if they want to be a cloud provider then they need the biggest most massive and efficient data centers. Or are we talking purely about small and medium data centers and colocs?
Sure, co-location is far too expensive relative to provisioning on-demand cloud infrastructure to justify for small and medium size businesses. And large businesses can set up their own data centers in-house. I could see some medium to large businesses buying these data centers just to avoid having to build their own from scratch.
The cloud will always suffer from the problem of physical custody of data. Where some data is too sensitive to trust to someone else's security even if that security is probably better than your own.
But for everything else the cloud is where it is at.
Data centers suffer from very rapid capital depreciation. Kryder's Law [pcmag.com] means your storage hardware
A datacenter is a physical facility, which provides floor space, power, heating/cooling, fire suppression, backup power generation, security, etc.
Kryder's Law is talking about particular types of equipment which is placed INTO a datacenter.
A lot of large companies are realizing that it's cheaper to lease/rent space from a company which specializes in the actual datacenter itself, than it is to try and do it all on their own. It also gives them the flexibility to move their gear to other facilities in response to changes in technology and network topology, etc. They don't have to spend a year and multiple millions of dollars to 'stand up' a new facility, they just find someone in the area and move in.
Datacenters are not going away. They aren't getting smaller. There are, in fact, more of them being built than every before. It's just the ownership and operation of the facility itself which is changing hands.
Remotely locating your servers, requires several ultra high speed internet connections per office space.
Otherwise you risk major amounts of downtime. Meanwhile local hosted servers can increase up-time, and increase available bandwidth on the cheap(wires in walls).
The price differential between cheap lower end business grade Inet connections, and high end fiber connections can be several thousand dollars a month or more. Also tack in the cost of very high end routers supporting large amounts of VPN traffic. Then let's not forget how much one must pay for the remote hosting sight..
The only time remote servers makes sense is when you have a application that faces a diversity of consumers already on the net.