Large or global businesses that rely on their websites to conduct business and present information to the public must include website hosting in their business plans. A term that’s abstract to many, “hosting” refers to where and how the website is stored on the Internet, and directly impacts the speed, reliability and uptime of the website. Let’s explore what types of large or global businesses are best served with an “Enterprise Hosting Solution” as a foundation for their website:
- Magento eCommerce businesses, particularly if customers are buying from around the world
- Businesses where the nature of the content is sensitive and must be strongly protected against hacking (example: chemical companies)
- Large corporations that must follow strict regulatory and legal guidelines regarding the content presented on their website (example: pharmaceutical companies)
- Companies who want their website to integrate with back-office ERP systems (example: retailers with a long history and extensive databases)
A Scalable Solution
As business owners learn more about the role of website hosting in their company’s evolving needs, we recommend they choose a hosting configuration that’s scalable, redundant and failure-resistant. The system must be architected to resist common datacenter issues, provide inherent disaster recovery, and provide a standardized high speed browsing experience for customers/users in all geographic regions.
Under the heading of “I didn’t know I needed it until I had it”, when business owners and/or their tech teams better understand the components of a global enterprise hosting solution, they can make more informed decisions about what’s right for their business. Here are the major factors to consider, details below:
- Amazon Web Services (AWS)
- Multi-region deployment
- Redundant and scalable architecture
- Redundant and scalable global database cluster
- Redundant replicating file system
- Latency-based DNS routing
- Single administrative login to manage all regions
Cloud-Based Hosting Platform – Speed and Reliability
AWS provides a plethora of Enterprise-Level cloud services typically billed by consumption rate. This allows for cost-effective increase of services with no upfront hardware costs and no long term contracts. Currently, AWS accounts for approximately 92% of all public cloud computing with the closest competitor (Microsoft Azure) accounting for approximately 4%. The remaining 4% of cloud computing is handled by thousands of other cloud computing providers. AWS’s client list boasts big name companies including Netflix, Spotify, Reddit, and of course its own eCommerce giant, amazon.com.
AWS offers specific training and certification for individuals who work on the system; we recommend businesses seek out hosting companies with programmers who have achieved this level of expertise.
See Hosting Options & Configurations for illustrations of hosting setups and Multi-Region Websites: Geolocation for eCommerce.
Multi-Region Deployment – Added Protection
A multi-region deployment is essential to the goal of maximum redundancy and resistance to catastrophic failure. Our global eCommerce solution is designed to continue functioning despite the loss of an entire region of the deployment.
- Scenario: A two-region deployment operates in Oregon, USA and Dublin, Ireland. In the event of a complete outage in Oregon (such as from a natural disaster), all services continue to operate in Ireland, with minimal or no interruption of website functionality. Customers who would typically be routed to the Oregon region are re-routed to the Dublin region. When the Oregon region services come back online, everything automatically synchronizes with the Dublin region (including automatic database synchronization) and regular service is resumed.
Redundant and Scalable Web Servers – Strength
The main horsepower of most Magento Enterprise stores is the web server. Operating your store on a single server limits the web server processes, memory and CPU that can be employed to serve your customers. In a global Magento architecture, each region has a set of web servers that reside behind a regional load balancer.
- Scenario: Each web server handles a share of the web traffic for that region. This allows the system to spread the workload between multiple resources and accommodate the loss of one or more web servers (including web servers down for maintenance work) without affecting availability of the website in that region. In addition, this makes it possible to scale up web server capacity to accommodate larger amounts of traffic (whether temporary or permanent) without having to move datacenters or purchase new hardware.
Redundant and Scalable Global Database Cluster – Data Safety
The database for a Magento storefront is constantly changing. While database backups are created several times per day, restoring to a backup (in the event of a problem) should be employed as the worst case scenario. In addition, because the store is global, the information in the database must be consistent across all regions. For this purpose, we recommend employing a global database cluster solution.
- Scenario: Each region has a primary and secondary database server. A proxy on the web servers submit database requests to these regional servers in a round-robin fashion, spreading the load evenly. Together across all regions, all database servers function as one unit. If you were to query data in one region, it would be the same data as you would find in any of the other regions the cluster resides in. Additional database servers can be added within a region to help decrease traffic on the individual nodes in that region. This can be implemented with zero downtime.
- Administration: For the administrator, this means consistent order and product data across all regions. Customers in one region see the same pricing information as customers in another part of the world; similarly, site administrators see the same order information, regardless of where in the world they are.
- More tech details: The database cluster is architected to survive the loss of one or more nodes in the cluster. Lost nodes that become available again simply synchronize themselves with the cluster and resume normal operations.
Redundant Replicating File System – Efficiency
In a global Enterprise Hosting System, the file system is located on a minimum of two servers per region. In the event of a failure of the primary file system in a region, the secondary file system becomes the primary. One region’s primary file server is selected to become the Primary file system for the global deployment. Any changes made to the primary file server are automatically replicated out to file servers across the world.
- Example: A product is added to the store using the admin interface located in Oregon where the primary file server resides. The product data is inserted into the database and the images associated with the product are written to the primary file server. The primary file server then replicates out these file system changes to all the file servers in the global network.
The combination of replicated database and replicated file server means that new products inserted in the primary region are virtually immediately available in all other regions.
Latency-based DNS Routing – Fast Browsing
One of the common problems with a single server eCommerce solution is the speed of the browsing experience for users based in countries outside the U.S. The faster the browsing experience, the better the sales are going to be, while a slow browsing experience could cause loss of sales and customers.
- Example: Your single server solution is located in Chicago, IL. Customers within the U.S. typically have a good shopping session, experiencing browsing latencies of around 50-80ms. However, customers coming from countries across the globe will have a much different experience. The Internet distance they would travel may cause browsing latency of 150-200ms, resulting in a slow browsing and purchasing experience.
The solution is Amazon’s Route53 Latency-based DNS routing. When a visitor goes to a website that has this set-up, Amazon’s DNS system identifies the closest regional deployment to their location and directs them there. This reduces browser-latency and gives the fastest browsing experience. For example, a user located in South Korea would be directed to the deployment in Asia, rather than the U.S., and thus experience quicker browsing.
Single Administrative Login to Manage All Regions – Convenience
A multi-region deployment of Magento Enterprise doesn’t mean you have to maintain several administrative interfaces. The system can be designed so that administrative updates are managed in one region and the rest of the regions are synchronized with those changes. Therefore, if you have three regions in your deployment, you won’t need to check every admin interface, only the primary one you select.
As you can tell, there are many complex factors that play into determining the best enterprise hosting solution for a company. The architecture can roll out over time, allowing the system and the people behind it time to address all the nuances involved.
Get a FREE eCommerce Business Assessment – learn how you can manage orders, product data, customer information, and accounting processes in one seamless system.
Additional Resources
Cloud Computing and Securing Your Data
Why Host Your Magento Store on the Amazon Cloud
Hosting a Magento Website on the Amazon Cloud
Managed Hosting Overview
Managed Hosting Configurations & Options
Multi-Region Websites: Geolocation for eCommerce
September 6, 2016
All Articles