This was Amazon's first developer conference - so I thought it would be important to spend some observation time and thought cycles on what they presented and announced in the keynotes around what I believe is the most dominant cloud infrastructure out there. 1st post is about Andy Jassy’s keynote, 2nd will be on Werner Vogels.
Andy Jassy's keynote on Tuesday started with a customer overview - with Pinterest, Instagram, Dropbox, Foursquare and Sprotify to large enterprises like Shell, Samsung, Hitachi - and many government and academic institutions working now with AWS. It begs more and more the question - who does not use AWS?
Some landmark data points are:
AWS has expanded to 9 regions and 25 availability zones (aka data centers) with 1 in the West, 2 on the East Coast of the US, one in Europe, one in Brazil, one in Tokyo, one in Singapore and the newest one in Australia. And one more, the 'gov cloud' for government in the US. The content delivery network now has 38 'edge' sites.
What amazes me the most is the variety of use cases - from media and streaming apps, to conventional business apps, consumer apps and even to games. That generic hardware can support these very different applications, without more specialization of the configuration is quite surprising (but more on the new instance types soon).
It was impressive to see how AWS helped JPL to stream the Mars Rover landing earlier this year. Theatrical how the throughput needed to scale up as the rover came closer and closer to Mars' surface. Ironic that the JPL website - built on traditional web infrastructure crumbled from the requests by eager viewers - but the AWS streaming instance kept up, serving up to 80 GBit / sec. I remember also impressive numbers from Google App Engine powering the Royal Wedding in the UK last year - so it looks like streaming events are the the true scalability tests of cloud infrastructures.
A true point in how Amazon's infrastructure gets better and cheaper is that Amazon has passed 26 price reductions to their AWS customer. And good plug to state that S3 will be about 25% cheaper to be used. Jassi also pointed out how capacity planning is less crucial thanks to AWS and showed some impressive instance delivery statistics. The combination of all these features allows for more experiments, more options to test out new innovative ideas. Moreover he made a good point that by AWS taking over the 'plumbing' (data centers, servers, networks etc), customers can focus on what is strategic - building great applications. And not surprisingly AWS gives you an excellent springboard to a worldwide platform.
As a public cloud vendor, Jassy of course needed to compare the public cloud to the private cloud, which according to Amazon does not even closely deliver the benefits of the public cloud. He then went on to quote some 'old guard technology' vendor statements, having some fun with them:
![]()
The Amazon stab against the 'old guard technology' vendors is that they are 60-80% gross margin businesses, that cannot think like Amazon, a high volume low profit margin business. Obviously this is the better company DNA to run a public cloud vendor.
The showcase of Netflix with CEO Reed Hastings on stage was impressive. Not only is it again a streaming business that taxes any infrastructure, but it's also a competitor of Amazon (Amazon Video Service) that trusts AWS to run its infrastructure on it. Not many other businesses would have even vaguely considered to run their key delivery IT infrastructure ... on that of a competitor. Hastings seemed to be very comfortable with that. How much Netflix could really have built this infrastructure themselves is of course another story. But today already 95% of Netflix runs on AWS. I was impressed with Hastings' cloud saviness - he called the current state of cloud 'assembly language phase' - comparing it to developers in the pre compiler age carefully having to assign registers to their programs. He also raised the ability to moving running live instance (with a mention of VMware - Jassi didn't look too amused) and sees that as the next stage for cloud platform providers. On the client side Hastings sees touch interfaces, lot's of personal computing power in the hands of people, but cloud assisted - Siri was the example. He also talked about the famous analytic problem to serve the right video content to their customers, which is the 'fundamental ranking' problem to the Netflix business.
The progression of functionality and features in AWS is remarkable - from a few dozens a few years back - to well over 150. The driver really is to make any application be able to run in AWS. It's worth mentioning that DynamoDB - only launched in January 2012 is the fastest growing AWS service.
So what is the next AWS service? Not surprisingly - almost the logical step given the existing capabilities AWS has for storage and analytics - this is Amazon redshift - a data warehousing service. Platforms range from 2 TB / 16 GB RAM nodes to 16 TB local disk and 128 GB RAM, you can configure up to 100 nodes (up to 1.6PB). Data (not surprisingly) is stored in columns and standard SQL as well as JDCB / ODBC queries are supported.
Amazon tested Redshift with their own traditional data warehouse (quite impressive - 32 nodes, 4.2TB of RAM, 1.6 PB disk (sic the size of the new instances?!) - and a several million investment) and was able to run it in two 16 TB / 128 GB RAM nodes (less space needed due to column storage) at $3.65 / hour or $32k p.a. - a pretty compelling business case. Drink you own champagne...
Looks like the tools vendors have realized this as Redshift supports Microstrategy, Jaspersoft, SAP Business Objects and Cognos through ODBC / JDBC / Postgres SQL access. Redshift is in preview mode today and will be officially launched in 2013.
Equally Amazon has done a lot of security certifications and adoptions of standards for AWS to improve businesses concerns around security. Nasdaq invested into FinQloud on AWS, adding credibility to the AWS security features.
Finally Amazon has done a lot of work to help customers move their applications to AWS, an effort also supported by a lot of partners in the ecosystem. The enterprise software group running on AWS is impressive - not only the 800 pound gorillas of SAP and Oracle / Peoplesoft - but also infor, Appian, Pega... oddly Microsoft Sharepoint was on the same slide.Sanjay Poonen did the honors for the segment - didn't hurt he and Andy were together at HBS. SAP now supports Business Objects, the mobility platform, Hana One and Business One on AWS.
And in closing Jassi mentioned that the Obama campaign used AWS to power their grass roots campaign. Well there AWS really helped to shape history and the near term future of the US.
myPOV: A promising start for the AWS reinvent conference. Not sure where the re-invention is – as this is an impressive progression of the AWS platform. As mentioned earlier – who does not move AWS in some sort of fashion – and can you really afford not to look at AWS in 2012?
More tomorrow when we hear from the CTO of AWS, Werner Vogels.
Andy Jassy's keynote on Tuesday started with a customer overview - with Pinterest, Instagram, Dropbox, Foursquare and Sprotify to large enterprises like Shell, Samsung, Hitachi - and many government and academic institutions working now with AWS. It begs more and more the question - who does not use AWS?
Some landmark data points are:
· S3 is now storing over 1T objects.
· The elastic Mapreduce services has run over 4M clusters in the last years.
· And Amazon grew from a $5.2B business in 2003 - and today on every single day AWS is adding enough server capacity to run a 5B business
AWS has expanded to 9 regions and 25 availability zones (aka data centers) with 1 in the West, 2 on the East Coast of the US, one in Europe, one in Brazil, one in Tokyo, one in Singapore and the newest one in Australia. And one more, the 'gov cloud' for government in the US. The content delivery network now has 38 'edge' sites.
What amazes me the most is the variety of use cases - from media and streaming apps, to conventional business apps, consumer apps and even to games. That generic hardware can support these very different applications, without more specialization of the configuration is quite surprising (but more on the new instance types soon).
It was impressive to see how AWS helped JPL to stream the Mars Rover landing earlier this year. Theatrical how the throughput needed to scale up as the rover came closer and closer to Mars' surface. Ironic that the JPL website - built on traditional web infrastructure crumbled from the requests by eager viewers - but the AWS streaming instance kept up, serving up to 80 GBit / sec. I remember also impressive numbers from Google App Engine powering the Royal Wedding in the UK last year - so it looks like streaming events are the the true scalability tests of cloud infrastructures.
A true point in how Amazon's infrastructure gets better and cheaper is that Amazon has passed 26 price reductions to their AWS customer. And good plug to state that S3 will be about 25% cheaper to be used. Jassi also pointed out how capacity planning is less crucial thanks to AWS and showed some impressive instance delivery statistics. The combination of all these features allows for more experiments, more options to test out new innovative ideas. Moreover he made a good point that by AWS taking over the 'plumbing' (data centers, servers, networks etc), customers can focus on what is strategic - building great applications. And not surprisingly AWS gives you an excellent springboard to a worldwide platform.
As a public cloud vendor, Jassy of course needed to compare the public cloud to the private cloud, which according to Amazon does not even closely deliver the benefits of the public cloud. He then went on to quote some 'old guard technology' vendor statements, having some fun with them:
Can you guess who the vendors are?
The Amazon stab against the 'old guard technology' vendors is that they are 60-80% gross margin businesses, that cannot think like Amazon, a high volume low profit margin business. Obviously this is the better company DNA to run a public cloud vendor.
The showcase of Netflix with CEO Reed Hastings on stage was impressive. Not only is it again a streaming business that taxes any infrastructure, but it's also a competitor of Amazon (Amazon Video Service) that trusts AWS to run its infrastructure on it. Not many other businesses would have even vaguely considered to run their key delivery IT infrastructure ... on that of a competitor. Hastings seemed to be very comfortable with that. How much Netflix could really have built this infrastructure themselves is of course another story. But today already 95% of Netflix runs on AWS. I was impressed with Hastings' cloud saviness - he called the current state of cloud 'assembly language phase' - comparing it to developers in the pre compiler age carefully having to assign registers to their programs. He also raised the ability to moving running live instance (with a mention of VMware - Jassi didn't look too amused) and sees that as the next stage for cloud platform providers. On the client side Hastings sees touch interfaces, lot's of personal computing power in the hands of people, but cloud assisted - Siri was the example. He also talked about the famous analytic problem to serve the right video content to their customers, which is the 'fundamental ranking' problem to the Netflix business.
The progression of functionality and features in AWS is remarkable - from a few dozens a few years back - to well over 150. The driver really is to make any application be able to run in AWS. It's worth mentioning that DynamoDB - only launched in January 2012 is the fastest growing AWS service.
So what is the next AWS service? Not surprisingly - almost the logical step given the existing capabilities AWS has for storage and analytics - this is Amazon redshift - a data warehousing service. Platforms range from 2 TB / 16 GB RAM nodes to 16 TB local disk and 128 GB RAM, you can configure up to 100 nodes (up to 1.6PB). Data (not surprisingly) is stored in columns and standard SQL as well as JDCB / ODBC queries are supported.
Amazon tested Redshift with their own traditional data warehouse (quite impressive - 32 nodes, 4.2TB of RAM, 1.6 PB disk (sic the size of the new instances?!) - and a several million investment) and was able to run it in two 16 TB / 128 GB RAM nodes (less space needed due to column storage) at $3.65 / hour or $32k p.a. - a pretty compelling business case. Drink you own champagne...
Looks like the tools vendors have realized this as Redshift supports Microstrategy, Jaspersoft, SAP Business Objects and Cognos through ODBC / JDBC / Postgres SQL access. Redshift is in preview mode today and will be officially launched in 2013.
Equally Amazon has done a lot of security certifications and adoptions of standards for AWS to improve businesses concerns around security. Nasdaq invested into FinQloud on AWS, adding credibility to the AWS security features.
Finally Amazon has done a lot of work to help customers move their applications to AWS, an effort also supported by a lot of partners in the ecosystem. The enterprise software group running on AWS is impressive - not only the 800 pound gorillas of SAP and Oracle / Peoplesoft - but also infor, Appian, Pega... oddly Microsoft Sharepoint was on the same slide.Sanjay Poonen did the honors for the segment - didn't hurt he and Andy were together at HBS. SAP now supports Business Objects, the mobility platform, Hana One and Business One on AWS.
And in closing Jassi mentioned that the Obama campaign used AWS to power their grass roots campaign. Well there AWS really helped to shape history and the near term future of the US.
myPOV: A promising start for the AWS reinvent conference. Not sure where the re-invention is – as this is an impressive progression of the AWS platform. As mentioned earlier – who does not move AWS in some sort of fashion – and can you really afford not to look at AWS in 2012?
More tomorrow when we hear from the CTO of AWS, Werner Vogels.