Technology leaders are being inundated with a flood of new cloud architectures, strategies and products – all guaranteed by vendors and various industry pundits to solve all of our database challenges. This seemingly endless array of public cloud based DBMS offerings can quickly become bewildering. One of the top questions our customers have is whether they should choose DBaaS or IaaS as their preferred database cloud architecture. This post is intended to peel back the veil of the 2 primary cloud based DBMS platforms by providing readers with our experiences with IaaS and DBaaS architectures.
One of the benefits of working for a remote DBA services provider is that our shop’s collective knowledge is not constrained by any one organization’s technology implementation. We have customers that have technology strategies that range from “bleeding edge” to “yesterday’s technology tomorrow.”
We know what products work and which ones don’t, what tech stack combinations play well together and what database technologies and features provide the most benefits for a given business or technical need. In addition, we are required to administer virtually every database feature you can think of for every product we support, and we work with dozens of cloud systems. This provides us with a wealth of knowledge that includes cloud strategies, technologies, architectures, product offerings and vendor-specific features.
DBaaS and IaaS Defined
Let’s continue our discussion by learning more about the two primary cloud architectures – Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS). Since we are talking about database management systems, let’s use the term Database-as-a-Service or DBaaS to define that architecture.
We all have experience with on-premises systems. We have to build the server rooms, provide heating, cooling, redundant connectivity and power. We are required to purchase, install and administer all of the components to provide the safest environment we can for our systems.
We are then required to buy the server hardware, install it and maintain it. When it breaks or we want to increase horsepower, we have to open the chassis and work on the server components. We add CPU, memory, disk – whatever we need. To perform those activities, we either have to take an outage or make plans to shift the system and workloads to another server to ensure availability. We also buy and administer the OS and DB software we need to run our database- driven applications.
In addition, we evaluate, buy, install and support all of the other products we need, which often includes monitoring, security, auditing and third-party reporting products. Look at the two stars in the the graphic above. We have to buy and support everything- both hardware and software.
Let’s move on to the cloud. Most IaaS and DBaaS environments are multi-tenant, which means we are sharing the vendor’s compute and storage architecture with other customers. In addition, depending on the architecture and vendor chosen, the system will vary in degrees of scalability, elasticity, automated administrative services and self-service.
The architecture that is the closest to on-premises is Infrastructure-as- a-Service, the architecture defined in the bottom center of our graphic. With Infrastructure-a- a-Service (IaaS), the vendor provides the compute and storage infrastructure and may offer some level of system maintenance activities. Customers have direct access to the cloud system, which includes both compute and storage components. Think of it as a server, more often than not, a virtual server in the cloud.
You don’t have to build your server support environment that provides air conditioning, light, multiple power providers, UPS systems, generators and redundant connections to the internet. All of those features are provided by the vendor, but IaaS customers will continue to maintain full ownership of their software stack's administration, including the operating system and database. Customers install and administer their software of choice on the Infrastructure-as-a-Service platform.
It’s important to note that depending on the IaaS provider and the offering chosen, customers are able to take advantage of the vendor’s features to reduce the time to support the environment. Microsoft Azure, for example, provides builds you can use to get a jumpstart on provisioning a new DB environment. However, you will need to tailor that generic build to meet your needs.
Now that we know that IaaS is pretty much a server in the cloud, let’s move on to PaaS, or in our case, DBaaS – Database-as-a-Service. DBaaS vendors provide all of the server environmental benefits that their IaaS counterparts do.
DBaaS providers increase their level of control and responsibility by assuming ownership of the operating system and database software as well as the hardware. DBaaS customers perform little to no operating system and database software administration. The vendors will be constantly enhancing their architecture’s automation capabilities to further reduce the human labor involved to administer their environments.
The DBaaS vendors also make modifications to their database software for two reasons:
#1-is to ensure their product will work in their shared cloud environment #2- to leverage the benefits that the cloud, and their architecture, inherently provides
Geographic data redundancy would be an example as it allows customers to leverage the cloud to more easily create DR and HA systems. It’s important to note that as we discuss IaaS and DBaaS architectures, there can be a lot of variations in the vendor offerings.
Comparing On-Premises, IaaS and DBaaS Architectures
Each environment, on-premises, IaaS and DBaaS, has strengths and weaknesses that are inherent to their architectures. I’ve provided this comparison information for the 3 DB environments shops can choose from – on-premises, the database running on IaaS and the DBaaS offering. Each environment has pros and cons, benefits and drawbacks.
IaaS allows customers to maintain tighter administrative control over their environment. They can also more easily leverage their favorite internal third-party products on IaaS systems than they can with DBaaS. IaaS is just a server that is provided to you over the cloud.
Many of the third-party tools that include monitoring, security, application development, auditing - can be challenging to integrate into a DBaaS architecture because of the modifications the vendors make to their systems. All of the DBaaS vendors provide monitoring tools. Some vendors, like Amazon, charge extra if the customer wants a more robust monitoring solution than what is offered in their base package. In general, DBaaS monitoring tools aren’t as robust as their on-premises counterparts, but they are catching up very quickly.
If the DBaaS provider determines that its own internal underlying software, OS or database needs a critical availability, security or performance patch, shops may not have a choice on its implementation. If that patch requires an outage, the customer will need to schedule that outage and oftentimes before a certain date. DBaaS offerings allow customers to more easily configure complex architectures such as high availability and disaster recovery. All of the major DBaaS providers offer geo data redundancy.
Leveraging DBaaS Environments to Reduce DBA Labor Costs
Migrating to a DBaaS environment does reduce the amount of time a DBA spends administering the database environment, but it doesn’t reduce that administrative time to 0. Customers will experience the most significant time savings in OS administration and hardware support.
DBAs do spend time installing, patching and upgrading the DBMS software as well as tuning the environment and setting up and monitoring maintenance and backup utilities. Many of these administrative tasks can be provided by the DBaaS vendor depending on the vendor and offering selected.
The majority of a DBA’s time is spent working within the database systems themselves. DBAs build schemas, grant security, assist developers with SQL and procedural program tuning, provide advice, enforce business logic using database features, tune application workloads and debug issues. DBaaS vendors don’t provide these services as part of their basic package. There are literally dozens (and dozens) of administrative activities that DBAs are required to perform in DBaaS environments.
It’s important to remember that although the vendor may provide the mechanisms and processes to automate administrative activities, this also does not reduce DBA support responsibilities for them to 0. Personnel may need to configure how they are to be performed and when they are scheduled. All of the major DBaaS vendors provide patching, maintenance utilities and automated backup processes, but it is up to the DBA to review scheduling, configure custom schedules and monitor the execution for all automated tasks.
An increasingly competitive DBaaS marketplace forces all vendors to maximize their product’s inherent feature set. Constant innovation and integration of new features that differentiate their product from other vendors is an absolute requirement for their continued competitive survival.
During my tenure in the IT field, I’ve found the following equation to be true:
New Features + New Functionality + New Technologies + New Architectures +New Business Requirements = Increased IT Support Complexity
As the cloud vendors add features to their DBaaS offerings, the products will increase in complexity. The features may automate some of the administrators’ support activities but, at a global level, DBAs with a high-level of technical knowledge will continue to be needed to support DBaaS environments.
Here’s a quick example. Most cloud vendors provide features that allow administrators to more easily configure HA and DR environments, but it still requires an extensive knowledge of these technologies and business requirements to configure, implement, administer and monitor. The activities may be performed much faster in DBaaS environments than on-premises systems, but that foundational base of knowledge is still required.
Well-trained administrators will determine what regions are best to house the failover systems, work with developers and business users to select a failover strategy that meets their combined business, technical and budgetary needs, monitor the actual failover process, identify the root cause of the failover to ensure it doesn't recur and restore the system to its original configuration once the problem is corrected.
Impact on Existing Change Management Procedures and Documentation
Because cloud environments are different than on-premises systems, organizations leveraging these new architectures may have to update their existing change management processes and documentation. How long that takes and the number of techs needed for the project, depends on, once again, the cloud architecture chosen and how stringent the organization’s change management processes and documentation requirements are. Because DBaaS is administered much more differently than on-premises, it will have a greater impact than IaaS, which is just a server on the cloud.
Additional documentation that may also need to be modified includes: security, disaster recovery, monitoring, problem resolution, job scheduling, administrative best practices, repeatable processes, internal, industry specific, governmental regulatory compliance, etc…
The amount of documentation changes required will depend upon the breadth and depth of documentation your particular organization requires as a best practice. Once again, this is a greater impact for DBaaS.
Impact on Existing Toolsets
Shops migrating to DBaaS environments will need to identify all of the build, administration, monitoring and access tools that they use to interact with their on-premises databases. All shops usually have a couple of “must have” tools that are frequently used. Administrators will need to identify which of their organization’s existing toolsets will continue to work in the cloud – and which ones won’t. The popularity of the cloud is driving most software product vendors to make sure their offerings work with cloud systems, but it is not something that should be taken for granted.
RDX’s recommendation is to create a list and verify that the preferred tools will continue to work with the cloud versions of the database. The majority of the tools will work with IaaS (remember, it is just a server in the cloud), but for DBaaS, shops will need to perform the evaluation to determine if they can integrate and the level of effort needed to integrate them.
Comparing On-Premises DB Feaures to IaaS and DBaaS
We learned that IaaS allows us to install on-premises DB software in the cloud. There’s not much analysis required to verify that all of the features we leverage in our on-premises databases are available in the IaaS cloud. Before shops migrate their favorite on-premises databases to DBaaS environments, they will need to identify the list of DB features that are not available in the cloud DBaaS offerings.
Let’s compare on-premises SQL Server to the 2 leading cloud vendors, Amazon and Microsoft, both of which offer SQL Server DBaaS. Amazon’s SQL Server DBaaS offering doesn’t support PolyBase, stretch database, backing up to blob storage and importing data into the msdb database. Also, administrators can’t rename a database if it is used in an Amazon Mirroring deployment, and it doesn’t allow customers to increase storage on a SQL Server database. If the customer needs to increase the storage of a SQL Server DB Instance, they are required to back the database up, create a new DB instance with increased storageand then restore the databases into the new DB instance.
It doesn’t support SSIS, SSAS or SSRS, and it doesn’t have SQL Server server-level security roles for sysadmin, serveradmin, securityadmin, dbcreator and bulkadmin. It also doesn’t support database mail, maintenance plans, distributed queries, log shipping, change data capture, SQL Server Audit or bulk insert.
How about Azure SQL DB vs SQL Server on-premises? These are two DB products offered by the same vendor. Microsoft’s Azure DBaaS offering doesn’t support attaching a database, backup and restore statements, change data capture, database mail, database mirroring, database snapshots, extended stored procedures, filestream, linked servers, log shipping, resource governor, the profiler, SSIS, SSAS or SSRS.
Now instead of database mirroring and log shipping, it does provide Active Geodata-Replication, which could be a better alternative for many customers – but its different! I’m not stating that these environments aren’t as effective as on-premises, but they have different features that we all need to be aware of.
Recent RDX customer surveys have shown that although most of our clients have defined their high-level cloud migration strategies, they are continuing to evaluate and compare IaaS and DBaaS platforms. Some of our smaller and mid-sized customers have decided to go “all in” and migrate all of their systems to the cloud, but the majority of our customers has stated that they will execute a “best fit” strategy which includes implementing new databases that use on-premises platforms as well as IaaS and DBaaS architectures.
In addition to choosing the most appropriate cloud architecture for a given database-driven application, there are a host of other considerations that must be evaluated when migrating to the cloud. Will you need to transfer data into and out of that cloud DB environment? How much data? Does the database being migrated depend on data from other on-premises systems? How will you ensure that the database and data transfers are secure? What level of application changes are you comfortable with? Are you permitted to have a downtime for the migration or is it required to be a “flip the switch” process? This is just a quick sampling. An entire article could be written on all of the issues that must be considered and evaluated when migrating to cloud architectures.
RDX has successfully converted dozens of on-premises systems to the cloud and changed DB products along the way. RDX offers a wide range of cloud DB services – from strategic analysis and architecture design to migration and ongoing support. If you would like our assistance, please visit our Cloud Services page for more info.