GET Amazon Web Services Infrastructure Part 1 / ua-hosting.company Blog / Sudo Null IT News FREE

The rows of servers interior the Amazon data center

After cloud computing emerged as a new paradigm and at long last became a separate area, Amazon was able to quickly become a leader therein field. The establish of Amazon Web Services in 2006 (just think about it, almost 10 years ago!) Allowed the retailer to become the largest actor in the securities industry, with a market apportion of $ 6 billion.

Over clock time, Amazon's cloud services began to serve tens and hundreds of thousands of customers (now more than a million). Accordingly, the uptime of services is very blistering, and even a small of downtime canful cost the company's customers very expensive. Not so long-wool agone, a failure nevertheless happened, and Netflix, Reddit, Touchwood, IMdB, and many other services suffered as a result. All this happened owed to a failure in a information center set in Virginia, USA. Now we offer to perplex familiar with the entire infrastructure of the company, describing its approximate geography and capabilities.

So, Amazon now manages at to the lowest degree 30 data centers of its ball-shaped network, and some other 10-15 will soon live well-stacked or are at the project stage. Unluckily, the company does not disclose the laden scheme of its infrastructure, but reported to indirect data, experts conclude that only in the USA the come content of the company's DC is about 600 MW.

According to Gartner, Amazon Net Services's computing top executive is quintuplet multiplication the total computing power of 14 other cloud providers.

Chess opening the veil of secrecy

Virago has been reluctant to talk about data centers since the coming of Amazon River Web Services, reporting a good deal less information about its infrastructure than unusual companies, such atomic number 3 Google, Facebook and Microsoft. Accurate, in the endmost hardly a years the billet has changed a trifle - meridian managers are already more willing to talk some the company's data centers.

"We are often asked questions some the physical substructure of Amazon World Wide Web Services. We've never said some about this area, and now we want to open the obliterate of secrecy about our network and information centers, "said Werner Vogels, Subject field Director and Vice President of Amazon Web Services at the July AWS To in Tel Aviv .

The independent goal of such meetings is to avail developers understand the Amazon philosophy regarding dapple base, as well as learn more about the system uptime and its dependability. So, the entire base is divided into 11 regions, each contains a cluster of data centers. There are several Availableness Zones in each region that provide customers with the ability to duplicate their services or mirror to avoid downtime. True, recent AWS infrastructure failures demo that here the Amazon squad could have worked more thoroughly.

Platform investiture is growing

Amazon Web Services grew 81% in the last quarter, compared to the Lapplander period last yr. This does not mean that the entire infrastructure is growing at the same order, but here you crapper be sure that Amazon is constantly adding servers, data warehouses and data centers to its infrastructure.

"Regular, Amazon increases the capacity of its base so that it would be enough to provide Amazon's globose infrastructure when the company represented the administration with an annual revenue of $ 7 billion," said James Hamilton, vice president and tip engineer at Amazon. This is quite a evidential.

Amazon's data center development strategy is now supported the idea of ​​monetary value reduction. Aside the way, since the set in motion of Amazon River Web Services, the price of services has dropped 49 times.

Werner Vogels (Exposure: YouTube)

"We do a lot to decoct the be of our services. The margin of our line of work is small, and we are entertained to keep it at the up-to-the-minute flat. Nevertheless, we reduce the cost of services regularly, "Werner Vogels commented along the situation.

The cornerstone of Amazon's entire strategy is determining the best size of data centers. Reported to company representatives, most Amazon data centers include from 50 to 80 grand servers, with a one-person data center capacity of 25-30 MW. The company went to this figure for a age.

How big should the data revolve around be?

As the size of Virago data centers increases, the time value of failure of a singular data center also increases. Accordant to company experts, the data center, and then to speak, is an element of failure. And the larger the data center, the greater the bear upon on the size of the unsuccessful person IT can induce. That is why the company does non create information centers with a electrical capacity of more than 100 thousand servers, most of its data centers are small and capacity.

The second question arises - how many servers keep going Amazon Entanglement Services? The information provided by the vice-presidents of the company allows us to talk about the stripped number of 1.5 million. The level bes number of AWS servers premeditated away Platform is 5.6 meg.

Amazon leases the creation of a number of data center operators, including Digital Realty Intrust and Corporate Office Properties Trust. In the past, the company leased buildings, such every bit warehouses, and converted them into data centers. More recently, Amazpn decided to change its strategy, focusing on creating DCs from scratch. In Oregon, the troupe used off-the-shelf modular components to assemble a sheer information center.

Amazon's advantage is too the foundation of its possess energy substations. Here, the need to ensure the speed of ferment, kind of than the direction of operative expenses, is already more evident. The nest egg are minimal, but you can create a data center at a much faster pace.

Custom servers and data storages

At the earliest stages of the development of its dapple platform, Amazon River purchased equipment from the most advisable-familiar manufacturers. Amazon's main hardware provider was Rackable Systems. In 2008 alone, Amazon ordered Severs for $ 86 million from this caller, and a year earlier for $ 56 jillio.

But with the growth of infrastructure, the company began to produce its ain computer hardware for its data centers. This allows Amazon to fine-line its servers, data warehousing and network equipment, optimizing the performance of complete hardware piece reducing costs.

"Yes, we are creating our own servers. We can buy ready-made solutions, but they are very high-ticket, and not too optimized for our needs. Thence, we create our possess equipment. We worked put together with Intel in order to be able to manipulation conventional processors in high-performance modality. This, successively, allowed us to create customized server types for role for very specialized purposes, "aforesaid Vogels.

Image: James Hamilton

EC2 instances use just such servers, assembled on the basis of Xeon E5 processors, made using a 22-millimicron process technology using Haswell architecture. Moreover, the servers here are of different configurations, premeditated to perform various tasks.

According to Amazon experts, the company now knows how to build servers with a form that is optimal for a certain range of tasks, including supporting the military operation of certain software and services.

AWS uses its own software also as hardware to build a network base.

Speed ​​of faint and cloud

The "light speed element" plays an important character in the design of Virago's infrastructure.

"The almost informal way for customers to work is aside launching applications in a limited data center, and you give the sack ensure maximal dependability of the data center for this option, agreeing that an uptime of 99.9% is quite sufficient. But if you are creating a highly tried and true application OR service, cardinal data centers are already required to ensure its operation. Furthermore, the distance 'tween the data centers can be very large, and the signal path is quite long. Therefore, the creation of a straggly infrastructure, especially if the data centers are distant from from each one other, can be a difficult tax, "said Vogels.

The answer to this trouble may be Availability Zones: clusters of information centers within the region, allowing customers to run instances in varied isolated locations, thus avoiding the point of failure. If something happens to one instance, the application is supported aside another, in another Availability District. From each one region has 2 to 6 Availability Zones.

At the same time, the company ready-made Handiness Zones isolated from each other, but cozy enough to ensure minimal signal postponement on the network. According to company experts, the signal delay between the zones is usually 1-2 milliseconds. For comparison, the wait en passant information from New York to Los Angeles is 70 milliseconds.

"We distinct to place Availability Zones close together. But they should stock-still represent in different geographical regions, and connected to diametrical Department of Energy networks, additionally, glucinium at different heights above seagoing level, "says Hamilton.

---

In the next article, we'll blab ou more about the geography of Amazon's meshwork infrastructure.

DOWNLOAD HERE

GET Amazon Web Services Infrastructure Part 1 / ua-hosting.company Blog / Sudo Null IT News FREE

Posted by: hardycumigho.blogspot.com

Share this post