Who "leaves the clouds": Western IT startups who abandoned the virtual infrastructure
At the beginning of last year , the research of the IT infrastructure market from IDC was released. According to him and Gartner's estimates , about 62% of IT infrastructure costs would have to go on placing in the data center in 2017, and shifting the share of physical infrastructure to virtual became one of the emerging trends in 2018 as well.
We have already talked about startups that have moved to the cloud ( here and here ). Now came the turn of well-known companies that leave them or combine virtual and physical infrastructure. Let us see why they make such decisions.
/ photo skeeze CC
')
Dropbox. For 8 years since its launch, the startup has not used its own hardware to accommodate customer data. He worked for Amazon Web Services. In 2013, the company began the transition to its own infrastructure, and by 2016 successfully placed about 90% of the total data volume in it (more than 600 petabytes).
Instagram made an exodus from Amazon Web Services in 2013 (a year after the Facebook deal). At that time, about 200 million people used the service, and during “Instagration”, as Mark Zuckerberg called this operation, it took more than 20 billion photos to be transferred to the Facebook infrastructure. The migration took the company about a year, and to complete this project it was necessary to expand the team of IT infrastructure specialists from 8 to 20 people.
Whatsapp Last year, it became known that Facebook plans to transfer the computing power of another “asset”. The messenger was one of the largest SoftLayer customers who was not yet part of IBM at the time of the start of cooperation in 2009. According to sources, the migration of WhatsApp to Facebook infrastructure was supposed to start at the end of 2017.
Tapjoy , a major provider of technology for advertising and analytics, was for the most part built on AWS. By the time the decision was made to transition to its infrastructure, it had 450 million active users and was a major customer of AWS. As the company developed, it explored the capabilities of SoftLayer, but decided to build its own infrastructure. Working with SoftLayer did not satisfy corporate requests, and part of the workload was moved to the Equinix capacity. As a result, Tapjoy chose the option of developing a hybrid infrastructure.
MemSQL started on a virtual infrastructure. At the start, the public cloud allowed the company to effectively deploy and test the product. In addition, participation in Y Combinator, brought a startup check for 10 thousand dollars for the use of AWS. Only two years after the project was founded, the management decided to switch to its own capacity due to a number of internal forecasts and reasons.
GitLab was one step away from switching to dedicated hardware 2 years ago. In terms of cost and performance, the solution looked efficient compared to the continued use of the virtual infrastructure. The company chose the CephFS file system, potentially scalable, but highly demanding on performance. As a result, the project has not switched to its infrastructure due to a number of reasons, including financial risks.
Zynga also changed her mind , but in her case it happened after giving up the virtual infrastructure. For most of its history, Zynga relied on AWS. In order to reduce costs and diversify risks after an IPO, the company began to invest in its own zCloud network. As of 2012, when the decision was made to switch to their capacities, 240 million players generated about 1.4 Pbytes of data, and only 20% of this volume was placed on AWS. By 2015, the company abandoned attempts to save on the construction of data centers and returned to the cloud.
Saving. The examples above are similar in one thing: at the initial stage, startups rely on the speed of development, and the use of virtual infrastructure gives teams the opportunity to focus on the business processes that make up the core of the company. When the bill goes to tens of millions of users, startups start thinking about complete independence, including financial independence - they find the necessary resources for the development of their own infrastructure.
Conflict of interest. As Dropbox evolves, the cloud corporation has evolved. Based on AWS, we managed to develop our own software and services that allow us to work efficiently with S3. Given its own technological development, the company actually found itself on the same market with Amazon, Google and Microsoft. Companies had a conflict of interest - from a strategic point of view, it was decided not to use third-party infrastructure as the main repository. By the way, a conflict of interest is manifested when working with partners. For example, Wal-Mart, which manages the world's largest retailer, requires its technology providers to leave AWS clouds — more recently, Amazon has competed with Wal-Mart.
Efficiency for global platforms. Information about the migration of WhatsApp to Data Centers Facebook did not come as a surprise. The company attaches all acquired services to its network. In the case of Instagram, this step has become a way to more effectively connect to a wide range of computing tools that help manage Facebook. In addition, buying Instagram meant working closely with Facebook, and the team chose to integrate Instagram with the Facebook infrastructure.
/ photo Maurizio Pesce CC
Savings in capital costs. A classic example of working on your own "hardware". Nick Craver (Nick Craver), software developer and system administrator of the company, told how the architecture of Stack Overflow. Not every company is ready to invest in such development and maintenance without confirming the business model and appropriate scaling.
Reduced fixed costs. First of all, it concerns personnel costs. 88% of cloud service users indicate cost savings. 60% of respondents said that the cloud allows the team to focus on strategic issues and not spend resources on infrastructure maintenance. Virtual infrastructure allows you to use only the necessary resources at a particular time - companies do not have to pay for unused capacity.
Risk diversification. A typical medium-sized organization uses or experiments with an average of eight cloud platforms of various types and from different vendors. This is due to the fact that one cloud platform can cope better with some workloads, and in the case of others, another one can provide higher performance and economic efficiency. Relying on one platform - even your own - to a certain degree of risk. A hybrid approach helps balance the load and solve atypical tasks.
PS Here are some more materials from our corporate blog:
PPS Our Topical Guides:
We have already talked about startups that have moved to the cloud ( here and here ). Now came the turn of well-known companies that leave them or combine virtual and physical infrastructure. Let us see why they make such decisions.
/ photo skeeze CC
')
Briefly about the "background"
Dropbox. For 8 years since its launch, the startup has not used its own hardware to accommodate customer data. He worked for Amazon Web Services. In 2013, the company began the transition to its own infrastructure, and by 2016 successfully placed about 90% of the total data volume in it (more than 600 petabytes).
Instagram made an exodus from Amazon Web Services in 2013 (a year after the Facebook deal). At that time, about 200 million people used the service, and during “Instagration”, as Mark Zuckerberg called this operation, it took more than 20 billion photos to be transferred to the Facebook infrastructure. The migration took the company about a year, and to complete this project it was necessary to expand the team of IT infrastructure specialists from 8 to 20 people.
Whatsapp Last year, it became known that Facebook plans to transfer the computing power of another “asset”. The messenger was one of the largest SoftLayer customers who was not yet part of IBM at the time of the start of cooperation in 2009. According to sources, the migration of WhatsApp to Facebook infrastructure was supposed to start at the end of 2017.
Tapjoy , a major provider of technology for advertising and analytics, was for the most part built on AWS. By the time the decision was made to transition to its infrastructure, it had 450 million active users and was a major customer of AWS. As the company developed, it explored the capabilities of SoftLayer, but decided to build its own infrastructure. Working with SoftLayer did not satisfy corporate requests, and part of the workload was moved to the Equinix capacity. As a result, Tapjoy chose the option of developing a hybrid infrastructure.
MemSQL started on a virtual infrastructure. At the start, the public cloud allowed the company to effectively deploy and test the product. In addition, participation in Y Combinator, brought a startup check for 10 thousand dollars for the use of AWS. Only two years after the project was founded, the management decided to switch to its own capacity due to a number of internal forecasts and reasons.
GitLab was one step away from switching to dedicated hardware 2 years ago. In terms of cost and performance, the solution looked efficient compared to the continued use of the virtual infrastructure. The company chose the CephFS file system, potentially scalable, but highly demanding on performance. As a result, the project has not switched to its infrastructure due to a number of reasons, including financial risks.
Zynga also changed her mind , but in her case it happened after giving up the virtual infrastructure. For most of its history, Zynga relied on AWS. In order to reduce costs and diversify risks after an IPO, the company began to invest in its own zCloud network. As of 2012, when the decision was made to switch to their capacities, 240 million players generated about 1.4 Pbytes of data, and only 20% of this volume was placed on AWS. By 2015, the company abandoned attempts to save on the construction of data centers and returned to the cloud.
Why companies leave the clouds
Saving. The examples above are similar in one thing: at the initial stage, startups rely on the speed of development, and the use of virtual infrastructure gives teams the opportunity to focus on the business processes that make up the core of the company. When the bill goes to tens of millions of users, startups start thinking about complete independence, including financial independence - they find the necessary resources for the development of their own infrastructure.
Conflict of interest. As Dropbox evolves, the cloud corporation has evolved. Based on AWS, we managed to develop our own software and services that allow us to work efficiently with S3. Given its own technological development, the company actually found itself on the same market with Amazon, Google and Microsoft. Companies had a conflict of interest - from a strategic point of view, it was decided not to use third-party infrastructure as the main repository. By the way, a conflict of interest is manifested when working with partners. For example, Wal-Mart, which manages the world's largest retailer, requires its technology providers to leave AWS clouds — more recently, Amazon has competed with Wal-Mart.
Efficiency for global platforms. Information about the migration of WhatsApp to Data Centers Facebook did not come as a surprise. The company attaches all acquired services to its network. In the case of Instagram, this step has become a way to more effectively connect to a wide range of computing tools that help manage Facebook. In addition, buying Instagram meant working closely with Facebook, and the team chose to integrate Instagram with the Facebook infrastructure.
/ photo Maurizio Pesce CC
Why do companies still rent clouds
Savings in capital costs. A classic example of working on your own "hardware". Nick Craver (Nick Craver), software developer and system administrator of the company, told how the architecture of Stack Overflow. Not every company is ready to invest in such development and maintenance without confirming the business model and appropriate scaling.
Reduced fixed costs. First of all, it concerns personnel costs. 88% of cloud service users indicate cost savings. 60% of respondents said that the cloud allows the team to focus on strategic issues and not spend resources on infrastructure maintenance. Virtual infrastructure allows you to use only the necessary resources at a particular time - companies do not have to pay for unused capacity.
Risk diversification. A typical medium-sized organization uses or experiments with an average of eight cloud platforms of various types and from different vendors. This is due to the fact that one cloud platform can cope better with some workloads, and in the case of others, another one can provide higher performance and economic efficiency. Relying on one platform - even your own - to a certain degree of risk. A hybrid approach helps balance the load and solve atypical tasks.
PS Here are some more materials from our corporate blog:
- Disaster recovery as a service: advantages and disadvantages
- IaaS: First steps after renting cloud infrastructure
- Best Practices for Migrating and Deploying IaaS Cloud Infrastructure
PPS Our Topical Guides:
Source: https://habr.com/ru/post/346070/
All Articles