Key Questions to Ask Before Selecting a Cloud Database
With so many databases and analytics providers on the market at various development stages, here are some key questions IT teams can ask before they zero in on a cloud database.
Businesses should carefully evaluate how cloud databases and analytics tools meet current needs while also balancing cost, complexity, performance, and flexibility. Steve Sarsfield, director of product marketing at Vertica, shares key questions to ask before selecting a cloud database.
If you’re planning to move your analytical workload to a single public cloud vendor, multiple cloud vendors, on-premises, or a hybrid infrastructure, the database you choose impacts costs significantly, productivity and business value. With so many databases and analytics providers on the market at various development stages, here are some key questions IT teams can ask before they zero in on a cloud database.
Can It Deploy Anywhere?
Some SaaS platforms require loading all data into one particular cloud. This locks the customer into one solution, not allowing the freedom to move to a different cloud easily or to take advantage of lower-cost computing when available. If a SaaS solution is labeled “cloud-native,” it may actually mean “cloud-only,” and it may not be able to run workloads in other places. In effect, you are locked into having to load data to one cloud and analyze it using a single engine. Billing convenience might be an advantage here. However, the business will only be limited to one type of cloud deployment.
See More: Top 5 Reasons To Migrate Databases to the Cloud
Can You Use Data from Anywhere?
External data and data lakes are becoming increasingly common in the enterprise, but analytics solutions can greatly vary in how they manage workloads and data storage. You want to be able to access data both within the database and externally. You want to give users access to more data sources, even if it’s not loaded into the database. The time you will spend loading the data onto the database and the amount of money an organization will pay while the data lives in the cloud can make a big difference. Storing all data in one database type is flawed as most modern solutions have combined the data warehouse and the data lake for analytics.
Is it Tunable and Elastic?
When users are trying to resolve slow-running queries, it’s typical for cloud-only databases to offer node-based optimization. If your queries are running slow, many cloud-based systems will add more nodes to spin up more compute power. That said, analytical workloads are not universal, and the performance of databases may be impacted by quarterly reports, a successful marketing campaign that caused more data to be generated, or poorly written queries.
This is why it is important to understand what options are available for speeding up queries. Look for systems that offer a massively parallel architecture (architecture may require manual sharding and special query modification to leverage the cluster); node scaling (scale nodes and also control node size or configuration); workload management (mapping of query resources like memory and CPU to specific query types or a particular set of users); separation of compute and storage (data is kept in object storage while compute nodes spin up to serve concurrency, backup, dashboards, and data science) and query optimization (query planners that figure out the best way to limit the data reads and memory needed to answer the query).
Does it Support Diverse Analytical User Roles?
As the cloud database becomes popular in the organization, be ready to support a wide range of requests from diverse users (business users, analysts, data scientists). Are the features offered, and at what costs?
One should always consider the depth of analytical functions that are on offer. Functions can include things like:
-
- Time series: SQL functions are built into the database to log data recorded over set intervals of time.
- Geospatial: SQL functions are based on latitude, longitude and elevation.
- Machine learning: The capability to train, manage, and deploy machine learning models.
- Alternate frameworks: Support for data science and additional languages outside of SQL.
Does it Help You Control Costs?
When moving analytics to the cloud, expenses can quickly spiral out of control, and enterprises can get locked into long-term contracts. Ensure that the solution allows users to spin down compute when not in use and has the ability to set clear limits on spending so that there are no surprises at the end of the month. IT teams should understand how the database auto-scales on long-running or complicated queries or when many concurrent workloads hit the system at once. If the database spins up extra nodes automatically, those additional nodes are automatically billed in monthly segments.
The ability to have shared storage is also critical because this allows multiple teams to use the same data without making copies. More copies equal more storage equals more money. Last but not least, many providers levy an egress fee, charged per megabyte, on data taken out of their platform. Be wary of platforms that will charge you to get data back out.
The fact is, business, technology, consumer expectations and the regulatory environment are evolving so rapidly that no one can predict the analytics and storage requirements for the future. Today we might move data to the cloud. Tomorrow, we might move it back on-prem, and on some days, we would like to use both. This is exactly why the database you choose should be flexible, scalable, and future-ready.
What are your key considerations for choosing the right cloud database? Tell us on Facebook, Twitter, and LinkedIn. We’d love to know!
MORE ON CLOUD DATABASE
