In a follow up to part 1, David Barker, founder and technical director at green colocation and connectivity supplier, 4D Data Centres, explains how to access externally stored data and how to deal with the most common issue when accessing storage over a network, latency.
How to access externally stored data
All externally stored data can usually be accessed over the public internet; however this might not always be the way you want to access it.
As a minimum any connection to your data that resides externally should be encrypted. In an ideal world this would be over an IPSEC VPN that will encrypt all your data while providing you with an extension of your local LAN. However some public cloud providers will only support SSL-based encryption via the web browser for accessing your files.
If you have selected a colocation provider or a cloud storage provider that is also able to provision connectivity back into your premises via a leased line or metro Ethernet-style service, then you have additional options that will be faster and more secure than a VPN over the internet.
Over these connections the provider can set up a private network (or vLAN) that will give you a direct connection back into your storage. Look for a supplier that can run at speeds from 10Mbps through to 10Gbps with fibre channel support. This allows you to really take advantage of your storage compared to a connection over the public internet that will be limited to the maximum speed of your connection. Depending on how the provider works it may also be possible to combine an internet feed along with a private connection back into your storage using the same fibre into your premises.
How to deal with latency issues
Latency is the time it takes for a packet of data to make the round-trip from your computer to the storage and back again, and is the biggest headache with externally stored data. If you have high latency then file access will run extremely slowly, and if the latency is high enough you will likely see data loss or corruption.
The level of latency expected should drive your choice of provider, storage type and location. For example, if you require synchronised replication across databases, you need to be able to access your storage with a latency of less than 5ms otherwise the databases will lose sync and the data will not be consistent across both locations.
Any connection to storage that uses the public internet, as in an IPSEC VPN, will be more susceptible to latency variations and at the same time there will be less that you are able to do about it.
To reduce latency you would need to work with your ISP or external storage provider in order to improve their routing and peering, or you need to move your data to another supplier that you know has a lower latency.
In order to control latency the best connections to use are the dedicated leased lines or metro Ethernet directly back into your storage. These will suffer far less variance as they are not reliant on multiple providers across the internet and have the ability to deploy Quality of Service (QoS) which guarantees the delivery of certain types of data (e.g. VoIP, iSCSI, etc.).
If you’re still suffering from latency issues, even on a leased line or metro Ethernet service, then it could well be the size of the connection versus the amount of data you’re trying to transmit. In that case the only ways to improve this are either to increase your connection size or deploy compression at both ends in order to try to make better use of the bandwidth you have available.
The decision to store your data externally isn’t an easy one. However, provided you manage the risks and ensure you have the connectivity in place to be able to use the data once it has been placed outside of your organisation it can be a very effective way of improving the availability and resilience of your storage.
Posted by David Barker, founder and technical director at green colocation and connectivity supplier, 4D Data Centres