Will the Mothership Land in Buffalo (aka DC's eating their young)

[DomainBop]

Data center owner/operator Centrilogic, owner of Dacentec and operator of a datacenter in Buffalo that counts ColoCrossing as one of its customers, is launching a cloud service called Mothership1  (news courtesy of serverian on LET and the WHIR). One of the locations it may launch in is its DC in Buffalo (yeah, the infamous mall).

Bracknell (UK) is next on the list, after that we will expand to one or more of our other data centers: Rochester, NY (USA), Buffalo, NY (USA), Mississauga, ON (Canada) and Hong Kong. http://lowendtalk.com/discussion/27011/mothership1-a-new-low-priced-cloud-from-centrilogic-and-incubaid/p2

My question is: what do you think the impact will be on smaller providers and pseudo datacenter operators (like CC) of datacenter owners/operators like CentriLogic, Quadranet, Choopa, Incero, etc. launching cloud services that are in direct competition with their customers?

My guess would be consolidation and an increase in deadpools/bankruptcies in the hosting industry.

 

[WebSearchingPro]

Pretty impressive, quite feature packed for the price. 

 

[blergh]

I don't think it will be that bad, if anything customers will demand more for less but quickly realize things don't quiet work that way.

 

[rds100]

There is one thing i can't understand - what's with these names?

 

[drmike]

This cloud / VPS offer will directly compete with CentriLogic's own customers.

Should be no surprise that ColoCrossing has been doing the same thing (selling VPS directly) for a few months.  Their prices aren't GVH giveaway competitive, but they are "affordable".

see: http://velocity-servers.com/

If CentriLogic had sense, they'd roll out the Dacentec offers to other DCs.  Until they do, they are taking the random stabbing at markets approach.

Big picture, all these companies competing in VPS segment are just indication of weakness / their market cap reached.

Eventually they'll wake up and see they've pissed off a big piece of their customer base who themselves are resellers of services (VPS most notably).   If they aren't interested in said market, then they should just raise their server rental prices and datacenter rates to more sustainable pricing.

 

[Conky]

Another week, another cloud provider launch. DigitalOcean, Vultr, Wabble, Mothership1. What next?

 

[SkylarM]

Could the name of the business be any worse? I thought Vultr was bad.

Makes sense though. Why should a DC rent a server at minimal profit with a pile of IPs when they can sell to VPS customers directly and profit tons per IP they sell? Will likely see more of it happen with IPv4 shortage. Colo/dedi rentals with IPs becomes far less profitable when IPs are of limited supply.

 

[drmike]

Another week, another cloud provider launch. DigitalOcean, Vultr, Wabble, Mothership1. What next?

All of these "brands" have entirely retarded names.  Nothing catchy about any of them, nothing relative either.   Enough of them are retarded spelling hipster plays.

Big picture, I can't see anyone paying an invoice to any of these names 5 years out.  They will fail or come to their senses and rename to something proper and marketable.

DigitalOcean is the semi-out of the group.  At least they didn't go calling themselves DigiOcn or something bastardized.  At least they managed to spell two words properly.

 

[DomainBop]

DigitalOcean is the semi-out of the group.  At least they didn't go calling themselves DigiOcn or something bastardized.  At least they managed to spell two words properly.

They just took a name that had already been used and thrown away in the '90's...

The original DO had 20 patents on 802.11 wireless and developed what was probably the world's first smartphone

http://en.wikipedia.org/wiki/Digital_Ocean

Another week, another cloud provider launch. DigitalOcean, Vultr, Wabble, Mothership1. What next?

Limestone and Quadranet quietly launched Cloud platforms.  CalPop which previously only sold colo/dedicated added shared hosting, and the list of DC owner/operators entering into direct competition with their customers for VPS/shared hosting market share goes on.

I think the biggest negative impact will be on small "host in a box" providers who are currently forced to compete exclusively on price because they have nothing unique to offer customers.  The difference between a $5 Vultr/DO KVM offering and a $3 OpenVZ offering isn't that much, and for the extra $2 customers receive the peace of mind of knowing they're dealing with a real company.

 

[Francisco]

All of these "brands" have entirely retarded names.  Nothing catchy about any of them, nothing relative either.   Enough of them are retarded spelling hipster plays.

Big picture, I can't see anyone paying an invoice to any of these names 5 years out.  They will fail or come to their senses and rename to something proper and marketable.

DigitalOcean is the semi-out of the group.  At least they didn't go calling themselves DigiOcn or something bastardized.  At least they managed to spell two words properly.

All the run of the mill clouds are likely just to make more long term money

while dedi's try to fill IP allocations. Quadra was originally charging $15/month

for their 512MB cloud offering as noted here:

http://www.webhostingtalk.com/showthread.php?t=1357071&highlight=quadranet

They were also "only" offering a /25 of IP's on a dedi for $229/month:

http://www.webhostingtalk.com/showthread.php?p=9046014&highlight=quadranet

Now, you got $5/month clouds and a /23 per dedicated depending on who you know.

These providers are just trying to protect their long term assets. These providers

are booking on higher than average churn on JUST those dedi's (say they keep it for

6 months compared to indefinitely) just so they can rescue those resources they put

away.

Anyway, I derailed so let me get back on track.

On the topic of Mothership1, I was predicting this to happen as they've been slowly moving

that way. While they originally claimed no involvement in the BurstNet buyout, don't be

surprised if the current burstnet owners try to hauck it to them.

I could make some string of technicalities about how people would be hosting with burstnet

now if that happened but I'll leave that alone

Francisco

 

[matteob]

i'm only one that consider all thats provider just vps with advanced billing? My cloud definition is pretty different...

 

[drmike]

i'm only one that consider all thats provider just vps with advanced billing? My cloud definition is pretty different...

I expect all sorts of features to be a real cloud.  SAN storage or highly distributed storage with a real backplane (no stupid single gigabit connector OnApp style).

At base, should be self migrating instances across a cluster in a facility at-will.  Ideally, ability to move across DCs / geography.   Who ever heard of entirely stationary Clouds

 

[matteob]

SAN storage or highly distributed storage with a real backplane (no stupid single gigabit connector OnApp style).

All features you mentioned are included in onapp, Also OnAPP I.S. support bonding or vlan isolation that allow you to be fail tollerant. In real cloud enviroment, with onapp, you can bond on stack switches, or use multiple datastore nic channel in different non stacked switches or use different vlan in single switch.

With onapp you can keep data copy in different hypervisor or in different DC in a easy way.

 

[drmike]

All features you mentioned are included in onapp, Also OnAPP I.S. support bonding or vlan isolation that allow you to be fail tollerant. In real cloud enviroment, with onapp, you can bond on stack switches, or use multiple datastore nic channel in different non stacked switches or use different vlan in single switch.

With onapp you can keep data copy in different hypervisor or in different DC in a easy way.

So.... OnApp does bonding, which is for failed link protection, not bandwidth aggregation... (i.e. 4 1Gbps = 1Gbps speed not 4Gbps)... Correct me if I am wrong here...

Sounds like some hackery with Vlans in an attempt to aggregate bandwidth...

 

[Wintereise]

Onapp is just something over traditional linux, nothing should be stopping you from configuring bonding.ko to do 803.2ad.

 

[matteob]

So.... OnApp does bonding, which is for failed link protection, not bandwidth aggregation... (i.e. 4 1Gbps = 1Gbps speed not 4Gbps)... Correct me if I am wrong here...

Sounds like some hackery with Vlans in an attempt to aggregate bandwidth...

bonding is not related to onapp, but is a networking feature. Bonding is managed by switches vendor differently. For example on cisco, etherchannel permit you to aggregate port and push (on 2 ge port) to 2ge, only on multiple  transfers. You can choose the mode, like ip-to-ip port-to-port etc. Brocade permit a full roundrobin (best solution for onapp that had native linux RR configuration).

In cisco you had two choice:

- Bond two port for failover ip and consider port aggregate on diffeerent socket opened

- Put each port in isolated vlan and let linux-rr work for you.

Best solution is using 2x10Ge port in etherchannel. Cloud backend storage can push to 20ge (10ge on single socket is enough) and you are redundant. Using 10ge port avoid bottleneck and possible issues