Data in a Multi-Cloud World
Utilizing multiple clouds from multiple vendors with a mix of public and private clouds is now a way of life for most enterprises today. However, operating in a multi-cloud environment is not without its challenges. Data egress charges, incurred when data is transferred out of a cloud to another, are a particularly thorny aspect.
I sat down with Derek DeHaan, Former Director- Business Development, Cloud & Managed Services at OneNeck, to discuss the importance of knowing your data to better understand and manage cloud costs more efficiently.
Q: In terms of cloud, what is data ingress and data egress?
A: Data Ingress – data communications and network traffic that is originating from outside the host network and destined for a node in the host network. This includes all data transfers into a hyperscale public cloud provider form your internal data centers or users at the edge of the network accessing resources on the public cloud platform.
Data Egress – data communications and network traffic leaving the host network and traversing public internet or private connections and going to an external site/network/host. Downloading information or accessing a web page from a source/server that resides in the hyper scale public cloud platform would be classified as egress traffic.
Q: What are the different aspects of data transfer in the cloud specifically regions and zone?
A: Regions – inter and intra region transfers are charged for differently as well as data transfer among different countries. Typically providers will have multiple regions within a given country to help provide redundancy and in most cases inter region data transfers are not charged for. Intra region transfers are charged for, but usually at a much lower rate than intra-country transfers.
Availability zones – most providers have a minimum of two availability zones in a region and usually a handful of data centers in a given availability zone. This is provided to help achieve certain levels of resiliency and the specified # of 9’s you will be seeking depending on your application requirements.
Q: How are ingress and egress charges in the cloud typically billed?
A: In most cases egress traffic is billed on a utility style model, you pay for what you use or transfer from one point to another. In some cases the providers offer dedicated connections into the platforms and provide unlimited plans for transfer of data which eliminates the need for doing it based on what is actually utilized. Unless you are transferring large amounts of data continuously this is usually cost prohibitive to go the unlimited route, so most organizations end up getting billed per GB or TB of data transfer out. Ingress traffic in all providers is free as they love to have organizations put data into their platforms.
Q: What are the best practices for minimizing/ controlling data egress costs?
A: The most important thing to consider is be sure you have an understanding of your application dependencies and data flows among servers and the users. By understanding user access patterns and dependencies you can help determine the amount of data that may be leaving a hyper scale public cloud provider. This will help you estimate the amount you may be billed for egress charges. Archive type data is also a great fit for hyper scale public cloud and many companies first step into the cloud – as you don’t access the data very often, so the storage is cheap and there are usually very minimal egress fees when data retrieval is required.
Additional Resources: