In a recent announcement, Google has unveiled two new features for Cloud Memorystore, their fully managed Redis service. These features consist of the beta of import-export, allowing to import and export data from and to other Redis instances, as well as the general availability of Redis 4.0 support.
With Cloud Memorystore, users of Google Cloud Platform have access to a fully managed, in-memory, data store service built on Redis, fully compatible with its open-source equivalent. Consequently, it delivers features such as in-memory caching, the use of robust data structures, persistence, and replication (when working with the Standard tier of the service), and pub-sub capabilities. Furthermore, Cloud Memorystore allows to instantiate instances from 1 GB up to 300 GB and delivers network throughput up to 12 Gbps. The service competes with other Redis cloud providers, such as Amazon ElastiCache and Azure Cache for Redis, and allows for uncomplicated creation and administration of instances, as described by Jamie McGibbon, digital project coordinator at New England Patriots.
Instances can be created and configured within a few minutes with admins simply having to choose a few options in order to create an instance, such as the tier, capacity, and region. The estimated monthly cost of the instance will be displayed on the configuration screen, based on the settings that are selected. If an application happens to have a spike in traffic, creating a need for more storage, admins can easily scale the Redis instances to meet the increased demand. Admins can view metrics related to the instances using Google's Stackdriver monitoring service, which provides information related to network throughput, latency, memory usage, and commands per second, among others.
One of the features promised with the initial introduction of the service on May 2018, was import-export of RDB files into a Google Cloud Storage bucket, which has now entered beta status. Accordingly, the import side allows loading backup data from existing Redis instances into the service, while export creates an RDB file, the snapshot persistence format of Redis. As Cloud Memorystore uses the Redis Database Backup format, it allows for complete interoperability with other Redis suppliers. Moreover, both import and export employ a regional Cloud Storage bucket to store the backup file and allow for configuration through the portal as the source or destination.
Source: https://cloud.google.com/blog/products/databases/cloud-memorystore-adds-import-export-and-redis-4-0
It is essential to note the restrictions on the import and export behaviors, as described in the documentation.
- The Cloud Storage bucket holding the RDB import files need to be in the same region as the Cloud Memorystore instance and needs to be of the default storage class of Regional storage.
- While importing data, the instance is unavailable until the operation finishes.
- Data is overwritten on successful import or flushed when the operation is unsuccessful or canceled.
- Import of RDB files only supports the same or lower versions of Redis.
- During the export, the instance is available to read and write actions, but not for administrative operations.
- While exporting, there may be additional latency on these actions.
Additionally, Cloud Memorystore now entered general availability of Redis 4.0 support, bringing various new capabilities. This support includes improvements in caching, thanks to a new LFU algorithm, and a new modules system which provides an API allowing to extend Redis with dynamically loaded modules. Equally important, the 4.0 release introduces two new memory management improvements, consisting of memory introspection, providing more information about memory, and active memory defragmentation, allowing to defragment memory while the system is online.