Service Databases
Data infrastructure designed for high availability, with monitoring
For companies whose critical systems depend on databases: we design where and how those databases run, with redundancy, monitoring and a recovery plan, on Azure, in your own data center or both. The goal is to keep the business running when a server, a disk or the network fails.
Summary
What tends to go wrong
The database itself can be well tuned and the business can still stop: a degrading disk, a single server with no standby, an unreliable network, an alert nobody received. Many companies only discover how fragile their infrastructure is on the day they need it, and there is no runbook saying what to do if the main server goes down.
Who it is for
Companies whose critical systems depend on databases and who need to decide, or revisit, where and how those databases run: on their own servers, in the cloud or both, with what level of redundancy and what kind of oversight.
Deliverables
What gets delivered
- A map of your current data infrastructure, with single points of failure identified
- An availability architecture recommendation based on RPO and RTO agreed with the business
- A documented recovery plan with the steps to bring the database back online
- A review of access and auditing on the database servers
- A monitoring dashboard or report with the agreed metrics
- An analytics architecture design covering source, destination, frequency, access, estimated consumption and who runs it
Steps
How the engagement runs
Platform snapshot
Inventory of servers, storage, network, versions and jobs, with access agreed with IT.
Business targets
Agreeing with the business how much data can be lost and how quickly the system must recover.
Design and alternatives
A recommended architecture, with alternatives and the operational effort each one involves.
Tested rollout
Delivered in stages, with testing before each stage is signed off.
Oversight
Continuous monitoring and regular reviews, when a support agreement is in place.
Roles
Who takes care of what
Trustwise
Assessment, architecture design, agreed implementation, monitoring setup and documentation.
Your company
Set the business requirements, hold cloud contracts and licenses, provide access to the environment and approve change windows.
Engagement
Where to start
We start by reviewing your current infrastructure and the level of availability the business needs. That leads to a proposal for a project, such as adding redundancy or moving to a new environment, or for ongoing support with monitoring. Commercial terms are set out in the proposal.
Scope
What is included
-
Redundancy and recovery plan
A redundant architecture, such as replicas and failover, and a recovery plan in proportion to what the business can afford to lose.
-
Platform monitoring
Health indicators watched closely to identify signs of degradation and work through the fix with your team.
-
Server access and auditing
Control over who reaches servers and databases, protection against accidental deletion and attacks, and an audit trail.
-
Server, disk and network bottlenecks
Analysis of CPU, memory, disk latency and network to tell an infrastructure bottleneck apart from a database or application problem.
-
Azure, your own data center or both
Environments on Azure, in your own data center or a mix of both, protected with the same care in every scenario.
-
Analytics platforms
Architecture for BI and AI on-premises, in the cloud or hybrid: ingestion, availability, identity, cost and monitoring.
Risk signs
Signs that your data infrastructure is fragile
The weak spots are usually visible before the incident. Nobody is looking at them, though.
- A single database server with no standby ready to take over
- Nobody can say how long the business would be down if it failed
- Problems found through user complaints instead of alerts
- Disks close to capacity or with high latency at busy times
- One server administrator password shared among several people
- A long-standing question about moving the database to the cloud, never analyzed
Availability
How much downtime can your business take?
High availability is not a product you install; it is an engineering decision sized to the risk. The design starts from the RPO and RTO targets the business sets. With those in hand, we assess whether a well-tested backup and restore strategy is enough, whether a standby replica is worth keeping, or whether the case calls for an availability group with automatic failover. Each option carries different licensing, infrastructure and operating costs, and the recommendation sets those differences side by side.
Monitoring
What monitoring keeps an eye on
Monitoring is there to show the trend before the incident. Metrics and alert thresholds are agreed with your team; these are the points usually covered.
- Availability of the instance and database services
- CPU and memory usage, disk latency
- Growth of data and log files
- Replica health and synchronization lag
- Failed backup and maintenance jobs
- Denied login attempts and permission changes
Alert channels, hours of coverage and response times are defined in each proposal.
Cloud or on-premises
Azure, your own servers, or both
There is no single right answer. Moving the database to Azure takes hardware upkeep off your plate and makes redundancy easier, but it changes how you pay and calls for attention to network, latency and licensing. Keeping it in your own data center gives you full control, along with full responsibility. We compare the scenarios using data from your environment and, if you decide to move, the database migration itself is handled by our SQL Server management service.
Analytics platforms
Where your BI and AI data will run
Dashboards and AI pilots need a platform that receives data from source systems, stores it securely and answers queries without affecting production. The design covers on-premises, cloud and hybrid environments and makes clear who runs each part.
- Ingestion: which systems, how often and by which route the data arrives
- Availability and recovery of the analytics platform itself
- Identity and access: who can read each dataset
- Estimated consumption and tracking of the monthly cost
- Monitoring of data loads and a named owner for operations
Azure and Microsoft Fabric are considered when they suit the client's environment and licensing.
Technologies
Platforms and tools
- Microsoft SQL Server
- Microsoft Azure: virtual machines and managed databases
- On-premises servers, physical or virtualized
- Always On availability groups and replicas
- Microsoft Fabric and Azure data services, evaluated against the client's environment and licensing
FAQ
Questions we often hear
How is this different from your SQL Server management service?
SQL Server management deals with what happens inside the database: queries, indexes, jobs, permissions and versions. Data infrastructure deals with where it runs and how it stays up: servers, storage, network, redundancy and monitoring. Many companies use both, but each can be engaged on its own.
Who decides the level of availability?
The business. RPO and RTO targets are decisions for leadership and the teams that depend on the system, not for IT, and they drive the architecture and the cost of the solution. Our article on testing backup restores explains both targets; during the assessment we help translate them into numbers for your environment.
Does high availability replace backups?
No. A replica copies mistakes too: if someone drops a table by accident or ransomware encrypts the data, the damage reaches the replica within seconds. Backups with sensible retention and tested restores remain the last line of defense, even with redundancy in place.
Is monitoring around the clock?
Hours of coverage, alert channels and response times are set out in the proposal according to how critical each environment is. We would rather agree on this explicitly than advertise a standard that may not fit your situation.
Is it worth moving the database to the cloud?
It depends on the size of the environment, connectivity between sites, current licensing and how much time your team spends on hardware today. The assessment compares the options using real data from your environment, so the decision is not driven by market trends alone.
Does our infrastructure need to change to use BI or AI?
Not always. Sometimes it is enough to keep analytical queries away from production; in other cases a dedicated analytics platform pays off. The assessment compares the options with data from your environment, including estimated consumption and who will run each part.