Supercharging Your Digital Infrastructure using AWS Global Accelerator
Introduction
The widespread adoption of cloud services by enterprises worldwide has significantly enhanced their ability to provide fast services to end users. As demand for high availability and top-tier performance continues to grow, businesses are searching for advanced solutions to meet these needs. One such solution that has gained immense popularity is the Amazon Web Services (AWS) Global Accelerator, which is now deployed by over 20,000 customers.
Deploying the Global Accelerator can have a substantial impact on application availability and performance, making it an excellent choice for businesses looking to enhance their services. Its introduction has been a game-changer for companies serving both local and global users. However, despite its numerous benefits, some enterprises are still hesitant to implement the Accelerator due to a lack of understanding of its capabilities.
To fully leverage the power of AWS Global Accelerator, organizations must take the time to understand its features and how it can help address their performance and availability challenges. With its advanced capabilities, the Global Accelerator has the potential to revolutionize business operations, enabling organizations to offer exceptional services to customers. By embracing this cutting-edge solution, enterprises can stay ahead of the competition and unlock new growth opportunities.
In this article, we will explore how companies have benefited from using AWS Global Accelerator, showing you how to supercharge your digital infrastructure with this powerful tool.
AWS Global Accelerator
Global Accelerator enables user traffic to reach your application via Amazon's global network instead of the public internet. The reason behind this approach is quite compelling. User traffic enters through Amazon’s global network at multiple edge locations, known as Points of Presence (PoPs).
Point of Presence (PoPs): AWS has edge locations spread across 94 cities in 51 countries. As of 2024, there are 119 Points of Presence (PoPs) worldwide. These PoPs act as connection points that allow user traffic to enter AWS's global, congestion-free network. The expansion of these PoPs enables AWS Global Accelerator to provide faster and more reliable performance for end-users.
![Global-Accelerator_Infrastructure-Map_[7.2023]](https://d1.awsstatic.com/Global-Accelerator_Infrastructure-Map_2x7.2023.4c2a53c7c0d445df26718987c0b6a4ff3a05510c.jpg)
Key Features of Global Accelerator:
Continuous Monitoring:
Global Accelerator continuously monitors the network, and in the event of a network failure, it intelligently reroutes traffic to the healthiest application endpoint that offers the best performance.
Availability:
AWS Global Accelerator offers an impressive 99.995% availability, providing an additional 500 uptime minutes annually compared to 99.9% availability. This difference is crucial for businesses with critical applications, where every minute of uptime matters. AWS introduced this feature to ensure businesses can access the highest levels of uptime for their essential applications.
Simplicity in Multi-Region Deployment:
AWS Global Accelerator simplifies multi-region deployment by enabling you to deploy applications in multiple regions and route traffic to the optimal endpoint based on your performance and availability requirements.
Global Static IP Address:
Global Accelerator is a networking service that provides global static IP addresses, combined with the power of Anycast. It also supports "Bring Your Own IP" (BYOIP) for IPv4 address prefixes, or you can use an IP provided by Amazon.
DDoS Mitigation:
Global Accelerator actively defends against DDoS attacks when used in conjunction with AWS Shield Advanced. It shields your traffic from volumetric and protocol attacks at the edge while swiftly mitigating threats closer to the source.
High Performance and High Resilience:
AWS's global network, which is used for inter-region traffic, is built to be congestion-free. Both CloudFront and Global Accelerator utilize this network, which is equipped with multiple redundant 400GB links. Each edge PoP has multiple links to the backbone and several paths to the region. AWS uses automation to maximize resilience, continuously monitoring the network and public IP routes to quickly identify and resolve issues before they impact your application.
Security:
With Global Accelerator, you can place your endpoint in private subnets, ensuring that endpoints are not exposed to the public subnet. Global Accelerator offers two types of protection:
Layer 3/4 DDoS Protection: You automatically receive DDoS protection via AWS Shield, designed to provide robust defense against network-layer attacks.
Layer 7 DDoS Protection: When using AWS WAF with an application load balancer endpoint, you benefit from protection against application-layer DDoS attacks, which safeguards your website or application from malicious traffic.
For AWS Shield Advanced customers, additional security features are available, including custom mitigation options, 24/7 access to the Shield response team, and DDoS cost protection to cover scaling costs due to DDoS attack spikes on protected resources. These features offer enhanced protection against potential DDoS attacks, helping ensure the security and availability of your resources.
AWS Global Accelerator is a Level 4 Anycast IP service that supports both TCP and UDP protocols, providing customers with a global fixed entry point to their Application Load Balancers (ALB), Network Load Balancers (NLB), EC2 instances, or Elastic IPs, routing traffic over the AWS global network to the nearest AWS Region to the client using Anycast IP addresses from the AWS edge network.
Case Studies: Companies Benefiting from Global Accelerator
Created using canva : Reference by aws case studies
A. theTradeDesk
A software company specializing in advertising technology uses AWS Global Accelerator to process real-time bids worldwide while maintaining load balancing across multiple regions.
The company provides a SaaS platform and develops add-ons for connected services such as Radio, TV, and Internet.
The partnership between the software company and AWS enhances the efficiency of advertising technology.
The platform processes around 10 million messages per second from different parts of the world.
The messages are processed from 7 data centers located around the world.
Three of the data centers are located in Germany, Tokyo, and Singapore and are provided by AWS.
Why did they move to the cloud?
Earlier, moving their product to the cloud was both technically and financially impossible. When the company first started, it ended up using the physical infrastructure. However, they soon discovered a problem. Due to latency and the capacity of their workload, they needed to move to the cloud. They closely worked with the accounts team and found a solution using Global Accelerator.
Although Global Accelerator was not designed to be a load balancer, the company found it very helpful to use it as one. They were able to scale to 10 million messages per second with low latency, responding to requests within a tenth of a second, or about 100 milliseconds. The workload ranged from a million requests per second down to 3 million when stepping down to the data center.
After migrating its bidding workload to AWS, The Trade Desk was able to reduce the build-time for a new market to less than a week, down from up to six months, and production deployment time to about three weeks.
AWS Architecture Discussion
Global Accelerator improves theTradeDesk website and acts as a load balancer.
As discussed earlier, Global Accelerator has an anycast feature.
GA was designed to accelerate traffic between AWS endpoints using VPC peering over the Amazon backbone.
If someone wants to send a message from Virginia to Singapore, instead of going out over the internet in 20 hops, it can now just go over VPC peering, which also has an anycast feature.
VPC peering is fast, reliable, and fully supported by Amazon.
What type of software platform is it built upon?
- Their platform is primarily written in .NET using C# (C-Sharp), and the front-end is implemented with .NET applications written in C#.
Anycast IP
One IP with multiple targets
Anycast is a method where a group of servers share a single IP address. The source computer sends data to the server that is closest, which reduces delays and saves bandwidth, making things faster for users. This also improves availability. It's essential to note that the closest server may not always be the one nearest geographically.
There are various addressing methods such as Unicast, Multicast, and Broadcast that we will discuss in detail in the coming days. For now, let's see which method The Trade Desk uses.
This was great for the industry, as many clients would cherry-pick IPs from the DNS record and send all their traffic to one IP, causing an overload on the assigned load balancer. Anycast helped The Trade Desk spread the load across multiple IPs. The Trade Desk has a Global Accelerator with two IPs (for example, IP "A") that are routed to the same evenly weighted target group "A".
How are they routing their traffic?
Earlier, when they moved to the cloud, they forklifted their system. The system was not designed for microservices and containers. But now, they use 2 EC2 instances, which are equivalent to physical instances. Everything is optimized for the cloud.
If the request comes in by the bidder, then it hits global accelerator and global accelerator routes it to the bidders' running on EC2 instances.
The DNS record lists two IPs.
Bidder request messages and traffic will hit one of these IPs.
The IP then distributes the traffic to a global accelerator.
The global accelerator is a Layer 4 appliance.
The global accelerator is fully managed by AWS.
The auction will begin and one of the bidders will be chosen.
The chosen bidder will respond with a bid request.
Depending on the bid, the bidder will either win or lose the auction.
If the bidder wins, they will get the plan. If they lose, they will not get the plan.
When Global Accelerator is applied to EC2, it is distributed across multiple regions and sites because the accelerator functions as a load balancer across multiple sites. There is one Global Accelerator for each site. While it can support multiple instances, it is currently unable to route to the correct instances. For instance, the Tokyo Global Accelerator is only for Tokyo instances.
Other parts of Architecture:
Database: They prefer Aerospike as their high-performance NoSQL database, which is designed to handle large amounts of data with low latency and high reliability. The database is running on EC2 instances. To further optimize its performance, they utilize NVMe (Non-Volatile Memory Express) drives to pack it with data and ensure even lower latency.
Advantage of EC2: The company has benefited from EC2 instances. For instance, instead of having to stay in Singapore or Tokyo when they need to have more capacity, they can simply add a few EC2 instances. As a result, instances can scale much faster compared to physical data centers.
Why Aerospike DB?
When a bid request is received, it is usually for targeted advertising or routing traffic to a nearby location within 10-12 miles of the restaurant.
Only necessary information, such as non-identifiable or personal information if required, is stored in the in-memory database.
The bidder retrieves information from the in-memory database, which provides fast and low-latency access.
If there are no bids, no low latency, or no SLAs(Service Level Agreements), then the company will lose its customers, and their business will suffer a huge loss. Low latency and SLAs are key important factors in business.
Amazon S3 ( Simple Storage Service )
AWS S3 (Simple Storage Service) is a cloud-based object storage service that offers high scalability, durability, and security.
S3 can store and retrieve any amount of data from anywhere on the internet, making it a popular choice for a wide range of applications, from websites to big data analytics.
S3 offers various storage classes, including Standard, Infrequent Access, Glacier, and Intelligent Tiering, to optimize costs based on data access patterns and business needs.
Tradedesk uses AWS S3, which is backed by a big database that runs and stores petabytes of information in AWS based on US East 1. This database is none other than Vertica.
Vertica
Vertica is an analytics database designed for big data analytics and business intelligence.
It is a high-performance database that can handle large-scale data warehousing and complex queries.
The service is backed by S3, which is a storage service provided by AWS.
When additional storage capacity is required, there is no need to add more drives to the Vertica cluster.
Instead, a new bucket can be added to the S3 storage layer to provide instant storage for the Vertica cluster.
This makes it easy to scale up the storage layer without having to add more physical hardware to the system.
In summary, The TradeDesk's architecture leverages a variety of modern technologies and best practices, including the use of AWS services such as Global Accelerator, EC2, S3, and other services, to create a scalable and high-performance advertising platform. The use of these services allows The TradeDesk to efficiently handle real-time bidding, machine learning, and other demanding workloads, while providing the reliability and performance required to compete in the fast-paced world of digital advertising.
B. Sky Scanner
Skyscanner is a travel booking company owned by Trip.com Group.
It facilitates flight, hotel, and car rental bookings for more than 100 million people each month.
The company provides content in 30 different languages.
Skyscanner's mobile application has been downloaded over 100 million times.
Why AWS?
When they had to build and deploy their services at a weekend, AWS helped them grow faster.
Earlier, it was not an easy task. They had issues with latency, and they could not afford downtime for maintenance.
As we discussed earlier in the TradeDesk case study, latency, downtime, and SLA certificate problems in the real world can cause users to jump from one site to another, potentially to competing companies. To overcome these issues, Sky Scanner found a solution by using the cloud computing service provider AWS.
Why they Choose Global Accelerator ?
Why they Choose AWS Spot Instances ?
Traffic management with AWS Global Accelerator
The website has been deployed with AWS Global Accelerator to eliminate its single point of failure. In order to further strengthen the availability and performance of its application, GA provides users with a static IP address that acts as an entry point to their application endpoint in one or multiple AWS Regions.
Now that Sky Scanner is using AWS Global Accelerator, they can take advantage of the vast, congestion-free AWS global network."
Global Network?
The global network helped Skyscanner to route user traffic through the closest AWS edge location to its application endpoint. AWS Accelerator provided resilience to endpoint failure.
Wonder How?
Global Accelerator routes traffic to the optimal AWS endpoint based on the geographic location of the user and the health of the endpoint. This ensures that traffic is directed to the closest available endpoint, regardless of the AWS region. Global Accelerator also provides fast failover capabilities that can shift traffic in less than 30 seconds to a healthy endpoint if the primary endpoint becomes unavailable.
After using AWS Global Accelerator, the company is now able to see a decrease in response time from over 200 milliseconds to less than 4 milliseconds, resulting in a 98% improvement. This has been achieved as they are always able to respond from the closest AWS region.
The health checking and failover capabilities of AWS Global Accelerator were able to speed up deployment time by allowing the application to run on EC2 spot instances.
Thanks to the automated traffic routing and failover capabilities provided by AWS Global Accelerator, the company has eliminated the need to build and maintain its own failover solution. As a result, they are now able to deploy new features and updates to their website in a matter of days instead of months.
The failover process on AWS Global Accelerator is so fast that the company is no longer concerned about losing any region. Thanks to the rapid failover capabilities provided by the service, they can ensure minimal downtime and uninterrupted service for their customers, regardless of any potential regional outages.
As a result, they shifted entirely from EC2 on-demand instances to Spot instances. Not only did this help them save 70% of their monthly expenses, but it also covered the cost of running AWS Global Accelerator.
Since Global Accelerator takes care of tasks such as managing Spot instances and EC2, the developers at Skyscanner now have more time to allocate towards innovating new ideas and features to attract even more customers, instead of building the entire architecture themselves.
In conclusion, the adoption of AWS Global Accelerator has brought numerous benefits to Skyscanner, from speeding up deployment time by allowing the application to run on EC2 spot instances to eliminating the need to build and maintain their own failover solution. The fast failover process provided by Global Accelerator has also enabled the company to ensure minimal downtime and uninterrupted service for their customers, regardless of any potential regional outages. With the cost savings achieved through the use of Spot instances and the ability to offload infrastructure management to Global Accelerator, Skyscanner's developers now have more time to focus on innovating new ideas and features to enhance the customer experience.
C. AWS services powered by Global Accelerator
- Amazon Simple Storage Service (Amazon S3)
Amazon S3 provides a Multi-Region Access Point, which offers a single global endpoint for accessing a dataset that spans multiple S3 buckets in different regions.
2. AWS Management Console
When some one connect to AWS management console, their session is been accelerated by global accelerator
3. Amazon Chime SDK
The Amazon Chime SDK uses a global accelerator, and with it, builders can easily add real-time voice, video, and messaging capabilities powered by machine learning to their applications.
4. Amazon Elastic Container Registry
Amazon Elastic Container Registry uses Global Accelerator for ECR public.
5. AWS CloudFront
Amazon CloudFront uses global accelerator for TinyDNS on EC2.
6. Amazon New World
Amazon's New World uses global accelerator. It is a massively multiplayer online role-playing game developed by Amazon Games Orange County and published by Amazon Games.
7. AWS Site-to-Site VPN
If anyone uses a VPN, then Site-to-Site VPN has an option called 'accelerate' that makes use of AWS global acceleration.
8 . Amazon Alexa
Amazon Alexa recently started using Global Accelerate for traffic acceleration from certain countries. Since Alexa started using it, the Alexa team has observed a significant 17% decrease in user-perceived network latency.
If you try to notice how Alexa is faster when you ask questions, such as 'Who won the Oscars 2023 in India', you can see the difference in the faster response time."
Conclusion
In conclusion, AWS Global Accelerator is a powerful tool that can supercharge your digital infrastructure in many ways. Skyscanner, for example, was able to speed up deployment time by running their application on EC2 spot instances and utilizing the automated traffic routing and failover capabilities provided by Global Accelerator. This not only resulted in cost savings but also allowed their developers to allocate more time towards innovating new ideas and features.
Other AWS services such as Amazon S3, AWS Management Console, Amazon Chime SDK, Amazon Elastic Container Registry, AWS CloudFront, Amazon New World, AWS Site-to-Site VPN, and Amazon Alexa also utilize Global Accelerator to improve their performance and accelerate their traffic.
By adopting AWS Global Accelerator, you can eliminate the need to build and maintain your own failover solution and ensure minimal downtime and uninterrupted service for your customers, regardless of any potential regional outages. With its fast failover process, you can have peace of mind that your applications will remain available and reliable even during unexpected events.
Overall, AWS Global Accelerator is an excellent choice for businesses that need to improve their application's performance, reliability, and availability while minimizing infrastructure management costs. It's a versatile solution that can help you optimize your infrastructure and stay competitive in today's fast-paced digital landscape.
Thank you for taking the time to read my article on Global Accelerator. Your interest and engagement mean a great deal to me. I'm thrilled to hear that the information provided was helpful in understanding its power. I appreciate any feedback or updates you may have regarding the article.
🔜 In my upcoming article, I'll cover exciting topics like launching EC2 instances and using VNC-like appliances to transition from CLI to GUI interfaces.
💡 I'll also provide valuable insights on using the CLI to launch AWS instances.
I'm confident this article will be just as informative and engaging as the last one.
🙌 Thanks again for your support. Stay connected for the latest updates and insights.