DPM Q3FY20 - Managing Cross-Platform Database Environments

Are the DBAs who are responsible for managing relational database management systems also responsible for managing non-relational systems (such as NoSQL and Hadoop)? (Only respondents who had deployed Hadoop or NoSQL technology) Yes 68% No 32% Figure 3. Managing non-relational DBMS relatively stable. Also present are open source platforms that have been around for at least ten years, such as MySQL and PostgreSQL. Of note are the relatively new entrants like MongoDB, Redis and Elasticsearch, whose popularity is increasing. The in-and-out, up-and-down nature of these rankings reflects the facts of life for about two-thirds of DBAs. Besides, supporting more platforms also means being nudged from the relational world of the DBMS into the non-relational world, as shown in Figure 3. Obstacles and recommendations So, why is it difficult to manage multiple database platforms? Most DBAs consider themselves experts in one primary platform, the one with which they are most comfortable. When cutting their teeth on a new, secondary platform, they tend to think, “How do I do this on MongoDB and make it similar to the way I’m accustomed to doing it on SQL Server?” It’s a valid point of departure, but before long they trip up on the differences between the platforms and on their own gaps in knowledge, training or experience. For some tasks, they rely on mature tools that are either unavailable or unready for prime time on new platforms. Even with an unfamiliar database, every DBA knows the familiar, common questions to answer to keep any system properly supporting the application infrastructure and performing well: 3 “Are my database servers available, or as available as they need to be?” Good database administration means knowing immediately that a database server is down rather than relying on users to raise the alarm. “Do I have a proper backup and recovery strategy, and is it working?” No IT group wants to be caught flatfooted in case of a failure or disaster. “Do I have enough space on disk to store persistent and temp data?” Most databases don't function well when they run out of room. “Are there critical entries in the database server error logs?” Every database server writes to log files, even if the information they contain is sometimes inscrutable. They can be a valuable starting point for all kinds of inquiries, so smart DBAs know where they are and keep an eye on them. That list will vary among organizations, but it contains the baseline standards one needs to enforce. Once they’ve been met or exceeded, DBAs can move to the next level of questions about capabilities: “How does the database server handle locking, and how do I diagnose problems?” “How can I find out which users are logged in to the database server? What are they doing right now?” “How can I view the most resourceintensive operations?” “Can I schedule database maintenance jobs to make the system healthier?” In some cases, DBAs rely on mature tools for legacy platforms that are either absent or unready for prime time on new platforms. Figure 4. Foglight for Databases, global view The answers to those questions help DBAs make decisions about resource allocation. They can lead to diagnostics on common tasks and measures as simple as knowing how to kill a session that's hogging resources from the rest of the workload. Each platform has its own methods, tools and APIs for database monitoring, and the learning curve for them is steep because they differ from one another. 4 Then, with the firefighting and database health issues in hand, DBAs can turn to the topics that make databases so important to the business, like performance tuning/optimization, high availability and security. And while database monitoring seems like the best way to keep an eye on performance tuning, high availability and security, it often introduces more complexity. Each platform has its own methods, tools and APIs for database monitoring, and the learning curve for them is steep because they differ from one another. As noted above, some platforms are too new to have mature, useful monitoring tools. DBAs managing three database platforms may find themselves juggling three different sets of monitoring tools, each of which presents information in a different way with different UI. MANAGING MULTIPLE PLATFORMS WITH FOGLIGHT Foglight for Databases is the database monitoring solution from Quest Software that is designed for cross-platform environments, as shown in Figure 4. Foglight monitors traditional database platforms like Oracle, SQL Server and DB2, plus newer platforms, both open source and non-relational, like Sybase/ SAP Adaptive Server Enterprise (ASE), MySQL (including MariaDB, Percona and AWS Aurora), PostgreSQL (including EnterpriseDB), MongoDB (all editions) and Cassandra (open source Apache and commercial DataStax). Its cross-platform coverage centralizes monitoring under a single umbrella so that DBAs no longer need bounce among different tools. Foglight also simplifies the UI of examining the health of databases, regardless of platform, by enabling drill-down from one layer to the next through clicks. DBAs enjoy a uniform view of statistics on instance performance, database availability, disaster recovery preparedness, remaining storage and server resources (CPU, memory, disk latency). The tasks of database tuning and workload analysis become consistent across database platforms, making life easier for DBAs.