While this is good news, AWS really ought to work on (a) competitive pricing, and (b) flexibility in cpu/ram customization. Today, google compute instances cost less than half of similar AWS instances and allow customization of ram and cpu.
I'm curious to know how Google gets efficient usage of their hardware when people are choosing weird RAM/CPU combos. Do you know if they have released any information on this?
I migrated a droplet from Singapore to Bangalore on the 12th. It hosts a site that was averaging ~ 15k page views per day, predominantly Indian traffic (banking recruitment guide). Since then, the traffic has been rising steadily and yesterday crossed an all-time high of 26k page views. (No other changes made to the site)
does Google take server location into account when showing search results? edit: seems they might. This is a probable cause for the increase in your traffic
Microsoft Azure had already (Edit: actually three) datacentres in India:
India is Microsoft tech heavy (things change though).
Funny story. A few years back, I got a call from one of their Azure BizDev reps. I told them I was happy they supported Linux and the guy wanted to hang up so fast...lol!
Point of fact: An AWS region is actually more than one DC. Mumbai has two availability zones which implies at least two DCs.
Also, Azure India regions are not generally available:
"The India regions are available to volume licensing customers and partners with a local enrollment in India. The India regions will open to direct online Azure subscriptions in 2016."
Slightly OT but Azure is such a better product than AWS. I was really skeptical because Microsoft has a legacy of doing an absolute shit job at all their copycat products. I happened into some credit and gave it a try during some downtime and was extremely pleasantly surprised. It's as if somebody applied a 2016 interface and usability to AWS.
It seems you are just talking about the UI of the web interface to the services. I agree this is an important feature that AWS doesn't do so well with. But in the end it's also has pretty much nothing to do with the underlying functionality of the services themselves.
I think this might be a case of "beauty in the eye of the beer holder". Azure's web UI (both the old one and the new one) is an abomination of the sort I've come to expect from MSFT design teams.
This is great news. I work at a major ecommerce player in India and we use AWS extensively, mostly from Singapore region. There have been internal discussions multiple times to move to a provider within India to prevent downtimes in case of an undersea cable cut or to gain the extra few fractions of a second in page load times but we always chose AWS for the flexibility it provides. Been waiting for this for a while!
Looks like all those billions Jeff Bezos is investing in India are trickling into AWS infrastructure as well. This is good news for others; Singapore was their closest zone for Indian consumers. I always wanted an availability zone here in India.
us-east-1 and us-west-2 pricing is locked to each other. Typically the power differences are related to the cost of power and transit/peering, not the capital of building the facilities.
Northern Virginia and Oregon have (generally) cheap power and cheap connectivity to anywhere in the world, with plenty of peering opportunities. In a developing country or with more expensive infrastructure (Mumbai, Tokyo) costs will be higher to get power and pull in fiber to peering locations with enough capacity to supply a provider of Amazon's size.
This is the correct analysis. Power costs are high in India and its often unreliable, requiring many industries to have expensive generators in place as backup.
Anyone know what the AWS China pricing is like? There does not seem to be anything on the web sites for S3 or EC2. None of the calculator sites seems to have China in their list.
And maybe its time for the Asia region to be broken away from Pacific (but keep both). Mumbai associated with Pacific introduces dissonance.
May be related: if hosting in China, you need an ICP license. Can't just open a shop or instance (in general... if Amazon have a way of getting around this, would be interesting).
I'm really curious about the power infrastructure they use within their DC to guarantee availability. Even when I was in a very nice part of Delhi there were frequent brown outs.
Completely false. Across three factories in two states, I don't get "guaranteed" power. In fact, the only place I see basically uninterrupted power is in the nice parts of residential Bombay.
Mumbai has no power issue. In the last 10 years I have lived here, I have never witnessed load shedding. It is because last mile is served by private power companies namely Tata and Reliance power. This makes power slightly costlier but a hell lot more reliable. We also have underground cables to the last mile so there are no surges
I'm based in the Middle East and was really looking forward to this since announcement we use the Singapore region and currently get 105ms pings to our instances. Yet, how come I get 135ms pings to Mumbai despite it being MUCH closer (~1930km vs 5840km to Singapore) ?
Lol. I am in South India, and when gaming, i frequently get better ping to Singapore than other parts of India; Also performance varies widely depending on my ip and the time of the day.
The routing in local isps is atrocious. I thought the situation would be different for commercial connections but your experience seems to suggest otherwise.
6 39 ms 39 ms 39 ms ix-xe-9-0-1-0.tcore2.MLV-Mumbai.as6453.net [180.87.39.57]
7 40 ms 39 ms 39 ms if-ae-2-2.tcore1.MLV-Mumbai.as6453.net [180.87.38.1]
8 131 ms 123 ms 124 ms 180.87.38.6
9 141 ms 135 ms 134 ms 115.114.89.118.static-Mumbai.vsnl.net.in [115.114.89.118]
10 143 ms 136 ms 183 ms 52.95.66.176
11 139 ms 136 ms 136 ms 52.95.66.197
12 124 ms 124 ms 123 ms 52.95.67.208
13 * * * Request timed out.
14 * * * Request timed out.
15 * * * Request timed out.
16 136 ms 136 ms 136 ms <Instance IP>
Whereas for Singapore:
6 145 ms 154 ms 146 ms 38895.sgw.equinix.com [27.111.228.215]
7 94 ms 102 ms 92 ms 52.93.8.10
8 94 ms 92 ms 92 ms 52.93.8.29
9 106 ms 104 ms 104 ms 203.83.223.31
10 107 ms 104 ms 104 ms <Instance IP>
The primary consideration for DC location is where your customers are located. If a startup is focused on the US market, such a move would make no sense. So, as with anything engineering, it depends.
That would be a hard-coded list. hoodoof means something that'd dynamically tell you what existed, not just what happened to exist when your copy of the AWS library was released.
Today, AWS has announced the general availability of its new India region, which should come as no surprise to any avid cloud engineer. AWS is not a new concept to India with its set of early adopters harnessing and driving cloud usage in the country for many years. As with all regional expansions, AWS must develop a highly knowledgeable and skilled local technical community. To help drive this process, Cloud Academy is excited to announce an introductory offer for Indian residents to celebrate the launch of the AWS India region.
