Navigation

CTO Blog – Architecture of PeaSoup Hosting

At PeaSoup, our philosophy is to be transparent and open to our customers. So the first transparency blog will be about our architecture.

As CTO I’ve been involved in many technologies in the past, some good, some excellent and some ok ish. This experience of technologies helped us to design our hosting platform. Our main Data Center is in London Dock Lands at the Telecity group. We have 4 independent ISP links going into our racks to have no single point of failure when a line drops or under performs.

Our platform itself is based around VMware vCloud 5.5. vCloud is a cloud computing initiative from VMware which will allow customers to migrate work on demand from their “internal cloud” of cooperating VMware hypervisors to a remote cloud of VMware hypervisors. The goal of the initiative is to provide the power of cloud computing with the flexibility allowed by virtualisation.

VMware vCloud is a leading platform that is designed to provide cloud services to our customers. vCloud has come a long way since it came on the market 2008.

Today vCloud is more aiming to service providers to provide their services to their customers. Companies that are looking for internal VMware cloud solutions will over time start using VMware vCloud Automation Center also known as vCAC.

Within PeaSoup, vCloud sit on top of a vSphere 5.5 platform. Two clusters are created initially. One for the management cluster, which holds all the virtual machines needed to run the vCloud Services, and one for the Resource cluster, which is dedicated for our customers.

The clusters are configured with standard Fujitsu servers. Why Fujitsu? We’ve been doing our due diligence over the last months and Fujitsu is the actually the third largest IT provider in the world! You might not have heard of them so much, but lots of systems in the UK are actually running on their systems (DVLA, HRMC for example). Secondly, I find it really important that the product that we choose is reliable. Fujitsu servers are among the vendors with the highest reliability rate today. All their servers are build and tested in a single place before shipped to us. As important, is the performance of the servers. The chosen servers are known for their performance, in fact our server model was one of the first to appear on the VMware hardware compatibility list for vSAN and has a few world record server benchmarks!

The switching infrastructure consists of a Huawei 10G network with a capability of 1.28 Tbps per switch. The front end switching consist of a Brocade 1 GB network which is used for inbound and outbound external network traffic.

PeaSoup is different than other vCloud providers today as we are not using any shared storage. We are fully adopting VMware’s new vSAN technology, which means that we provide an excellent scale up and scale out service without the need to buy loads of expensive disk arrays to start with. And more important, this also helps to keep the cost down for our customers. Both our clusters are using vSAN to hosts the virtual machines.

So what is vSAN?

A lot of great articles have been written in the past, especially by Duncan Epping and Cormac Hogan. First of all, vSAN is not an appliance. vSAN is a software based distributed storage solution which is built directly in the VMware hypervisor as kernel modules. vSAN leverage local storage from the ESXi hosts which are part of vSphere cluster. Each ESXi node must have one SSD and maximum 7 magnetic spinning disks per disk group. Within vSAN these two types of disk will form a disk group. An ESXi host can have up to 5 disk groups. A minimum of three hosts is needed to create a vSAN cluster and within the vSAN cluster, a vSAN datastore will be created. The niche here is that its performance is second to none and the data protection is of a very high level. The vSAN data is always stored on a second node and as an extra addition the meta data is copied also to a third node. This means that if a disks fail in a hosts, the vm’s will fail over to a second hosts if this is a fatal disk failure, the third node will automatically be prepared as it already holds the meta data, and now also receives the data. When the disk failure is fixed, the disk group will be rebuilt automatically.

The great thing of vSAN is that when PeaSoup gets new customers, we will either add additional disks for a new disk group or add extra hosts to the cluster and the vSAN will automatically be expanded. More important, our service won’t have any performance degradation, in contrary. It will only get faster.

PeaSoup will use a very high standard MLC SSD in our vSAN which will be able to deliver more than 30.000 IOPS per disk. Over the next articles I will dive a bit deeper into IOPS.

Harold Buter

Chief Technology Officer.


Share

Comments

comments powered by Disqus