MIA to SEF is around 50ms on a good day. MPLS providers will set their SLA around that value, so it's a pretty good bet the Internet as a whole will have some induces latency. Don't forget all the regeneration of signal and hops along that airmles route.

Ask the high frequency traders, ad providers, or other brokers how important latency is and you'll get a different story. There's a good reason people colocate at 111 8th or other downtown locations in Manhattan then across the river in Secaucus NJ. Data centers are always going to be a) close to the companies that house CPE there; b) close to others they wish to communicate with (NYSE, NASDAQ, CME) and then close to the carrier hotels (1 Wilsire, 56 Marietta, etc), and backed up by SLA driven networks, not just Network Service Provider connections.

PUE is more achievable in homogeneous DC's that Google, Facebook or Apple run. It's harder to get a PUE under 1.8 when in a heterogeneous data centers. It's hard to convey to customers that their per-kWH cost isn't just a markup on the utility power, but everything else that goes along with cooling, maintainability, and DC operations.

I was looking for your comment before adding it myself. Totally agree that for desktops and laptops, DO.. NOT... TOUCH... MY... SCREEN... I've had other engineers poking the hell out of a Visio diagram on the screen and sales execs tapping various Powerpoint slides. Initially I'll make a comment such as "it's not a touch screen" or "do you happen to have a microfiber cloth?". Now if I see anyone starting to move toward the screen I'll place my hand between theirs and nicely ask them not to touch the screen.

Ironically, on my smartphone or iPad, fingerprints, smudges and such don't bug me at all. You can leave a salsa and chocolate trail and it's no big deal.

So I think if the laptop was used for standard productivity, no touchy my screeny. I'd like to try a hybrid for a bit (Windows 8) and see if that changes my mind.

Agreed. The interface is sometimes confusing, but the ability to pull in customer details and tie it to custom services shows off it's flexibility.

Internap's cloud platform, from the Voxel purchase, uses it for current and upcoming services.

And colo providers get driven down on cost as customers forget the previous catastrophic event. Every dollar spent enhancing a data center is passed along to the customer in some fashion. When your per-cab or per-foot pricing is higher than the competition--even if you can demonstrate more resiliency--it is a harder sell.

You can talk flood plains, gen sets, fuel locations and delivery schedules, CRAH and chiller redundancy, etc. As long as the SSAE16 or SOC 2 is current and there are no lingering memories of an "event" at the facility, it's going to come down to price. It's hard to help customers understand the basic capabilities of colo, let alone pricing differences. It mostly seems to come down to perception of the facility -- and price.

I would propose customers that have stringent RPO/RTO's investigate true geographic diversity. Not just for separation of footprints, but also to be disaster-diverse. Hurricane, tornado, power brown outs, earthquakes, etc.

They are downtown because all the fiber terminates there (111 8th, 60 Hudson, 75 Broad). They originate from the teletype days. Closest point to the main meeting places is a good place to build data centers. Hence Ashburn, London, Frankfurt, etc.

111 8th apparently has common fuel from the landlord (no individual tanks). This morning a valve failure starved out some gen sets.

It is still wild and wooly in downtown....

What's remarkable is the PUE factors Google, Facebook and Apple can get in their data centers. I stil think these are due to the homogenous nature of the equipment the place there, and the fact they don't have to worry about the multi-tenancy of commercial data centers. Middle of nowhere locations where things such as venting from the hot aisle are possible. In NYC, the 111 8th Avenue data centers are a good example of the constraints put on the various operators. Hopefully Google can help remediate that.

In regards to power and cooling, I still have a lot to learn about the tradeoffs in ease of use versus PUE. Good thread.

Approximately 100-200 BTC lost here due to wallet being lost. That was way back in the early days before the GPU computations started.

The regional stats aren't correct. In looking at the regional winner for Georgia, I see it's Verizon FiOS. That would be news to VzT (Verizon Telecom) since they have zero presence in Georgia (AT&T, formerly BellSouth territory). People in AL, TN, SC and NC would also agree. My guess is that the numbers for Tampa (LEC is VzT) destroyed the performance for the rest of the ISP's checked in the other states.

I would love to see a similar test performed, at a higher level of quality, for ISP providers in data centers.

FiOS (Verizon) or uVerse (AT&T) are wireline products. If you look at wireline growth in the 2-4% year over year versus wireless in 20%+ range, there isn't a big incentive for the telco's to invest in wireline upgrades. Look at FiOS - there are no new markets or expansions that Verizon wishes to do. Why? Because the capital invested in a wireline plant doesn't have nearly the same return on capital invested that upgrading a market to LTE does.

Granted wireless is saturated on the voice call component, but data is still a growing market. Ever wonder why the new share plans give you unlimited minutes and SMS? That's a gimmie compared to the hope you'll consume data (and pay for it).

The new cabal is the broadband providers such as Comcast/Charter/Cox looking to enhance and augment the Ethernet data markets while the telcos focus on wireless. Hence the recent purchase by Verizon for frequencies while promoting the cable consortium. I still think you'll see any carrier with wireless sell off their wireline business as focus on "mobility".

Most of the newer submarine cable systems have large amounts of participants in the consortium. It's not uncommon for Verizon, AT&T, Sprint, Global Crossing, Tata, BT, and others to be involved in these systems. My guess is that FB wants more control over the capacity of the system or to ensure capacity ends up where they need it, say China. Interesting landing points for China too.

This is strange though. You'd think FB would rather invest in the data centers near big pipes instead investing in transit itself. Maybe they just want to circuits installed in a more timely manner--I'm looking at you Big Red.

Economically it may be more efficient to donate money, but the personal effort and involvement goes a long way. And, you can do both. A local food pantry we assist has seen a 600% increase since late 2008. Besides time and personal donations bought at Wal-Mart or Target, the donations are used by the pantry to buy in bulk things such as masa harina, sugar and such, and then packaged for individual families. Tis the season to be giving....

They seem to help too (World Vision that is). In Haiti, besides UN and Catholic Relief Services (CRS) vehicles, World Vision is predominant, even in the Central Plateau. Doing the research on local, regional and global agencies goes a long way. Even if you focus on one particular area (Haiti, local homeless, your neighbors, etc.), giving a small portion to other groups is helpful.

Streaming not available via Tivo at this time. At least not on the Tive-HD.