There are countless ways to deploy a cluster. Here, we have listed some of the more common options, with a configurator at the end to help visualize what these options look like when they are combined in various ways.
A basic cluster consists of two physical servers. The front-end server is connected to the internet and provides visitors with access to a service, typically a webserver. Content is fed to the front-end by a dedicated back-end server hosting a database through a private network.
By separating your database and content delivery services onto two separate devices, each server is able to dedicate their full resources to their primary task, rather than sharing resources. In many single-server solutions, the web services are competing with the database for resources when under heavy use.
More efficient use of hardware
By separating your database and content delivery services onto two separate devices, each server is able to be tailored to the service it supports. Database servers can make use of a more efficient drive configuration, while saving cost on the webserver’s drives, where performant storage is generally less important.
Virtualization of your environment is done by turning your server into a hypervisor. This hypervisor allows for the creation of virtual machines (VMs) to perform dedicated tasks such as load balancing, caching, and web services.
Virtualizing your environment allows you to create separate VMs dedicated to each specific task. For example, a web server VM can be stripped of anything not directly related to serving websites.
Advanced server management
Virtualizing your environment allows you to take advantage of special tools, such as snapshotting, running backups, and resource pooling. These features allow you to balance your workloads, secure your data, and quickly reallocate resources where needed.
Hypervisor pooling is the practice of joining multiple hypervisors together to create a large, shared, virtual environment. Physical resources are shared and Virtual Machines (VMs) can migrate from hypervisor to hypervisor.
Hypervisor pools have more resources than a typical single hypervisor, allowing for a greater number of VMs to be supported. In addition, additional hypervisor nodes can be added in the future to scale out your environment as needed.
Hardware and application redundancy
By building your virtual environment on a pool of multiple physical servers, any individual server in the pool can fail without interrupting your entire infrastructure. Recovery can be accomplished without extended downtime by starting the virtual infrastructure on another hypervisor in the pool.
Incorporating a public cloud into your cluster results in what is known as a Hybrid Cloud.
Utilizing GigeNET’s Cloud VMs in your environment’s front-end is ideal for environments that require significant scalability, as you can add and remove VMs and resources to your cluster at will.
Take advantage of GigeNET Cloud's flexibility by adjusting the size and number of your VMs on-demand to meet the changing demands of your business. Don't pay for resources you don't need, but have them at the ready for when you do.
Superior Hardware Redundancy
Built upon a massive pool of state-of-the-art hardware in multiple data centers across the US, this opens the door for adding geo-redundancy to your applications.
Keep the data stored in your database synchronized across two or more physical servers in real-time.
Data replication allows you to store two physical copies of the same data to two or more systems. This allows an active / passive setup to mitigate data loss due to failure. Being able to promote the passive database in an outage leads to a massive improvement in the overall recovery time.
Multiple in-sync databases allow you to take advantage of split workloads. This allows you to write to the active node, while reading from the passive node. This can vastly increase the performance of your database overall.
Uplink your front-end servers to a 10 Gbps uplink to The Internet.
Improved Throughput and Latency
A 10 Gbps connection allows for approximately 10x more traffic to your website without congestion. In addition, you will also see improvements to your latency under load.
Connect all the servers in your cluster to a 10 Gbps private network.
Improved Throughput and Latency
A 10 Gbps connection allows for greatly increased traffic between your servers without congestion. In addition, you will also see improvements to your latency under load.
Improved access to advanced storage technologies
A 10 Gbps private network allows you to take full advantage of advanced storage technologies, such as Ceph and Gluster. The improved throughput prevents any network-based bottlenecks between your systems and their storage.
Often referred to as Hyperconverged Infrastructure (HCI). This approach to virtualization further virtualizes all elements of the hypervisor pool and allows for federated management.
Federated management allows you to manage your locally shared storage across all your hypervisors. Gain full access to your own storage without having to rely on 3rd party control. Avoid inconvenient maintenance schedules and scale it out as you grow.
Design Your Custom Cluster
Use our cluster configurator to visualize your new cluster by trying out various options.
There are countless solutions possible, so contact us to start the process of designing a custom cluster built for your exact needs.