ERP and databases

Administration
of MS SQL databases

The database behind an inventory system runs for years without complaints, yet its failure can stop sales and production. We look after it before a failure happens: backups, indexes, a maintenance plan and test restores.

Ask for a quote +48 887 831 588

Server racks in a server room 3 typesof database backup
MS SQL Server Backups and restores Query performance Indexes Permissions Monitoring Licences and limits Archiving Recovery plan
4areas: indexes, statistics, maintenance, backups
Windowfor maintenance work, usually overnight
24/7emergency support
Testrestore of backups in a separate environment

Four areas of MS SQL administration

MS SQL administration covers four areas: indexes, statistics, the maintenance plan and backups. Sometimes some of them have not been checked since the day the system was deployed.

Indexes become fragmented when there is no job that rebuilds them regularly. Outdated statistics make the engine choose poor execution plans, so the same query takes a second one time and a minute the next. The maintenance plan is often set up with a wizard at deployment and has not been reviewed since.

Another risky setup is a full daily backup saved to the same disk the database sits on. A disk failure or ransomware encrypting the data then destroys the database and the backup at the same time.

  • Rebuilding and reorganising indexes according to their fragmentation
  • Updating statistics so that query execution plans are accurate
  • A maintenance plan matched to the company's maintenance windows
  • Full, differential and transaction log backups on an agreed schedule
  • A backup stored away from the production server
  • Monitoring of free space, queues and long-running queries

Where slow operations come from

A report that “the system has slowed down” rarely concerns the whole system. Usually two or three operations are slow: month-end closing, generating a report, a stocktake. Everything else runs normally, which is why the problem is often reported only months later.

We start by measuring. We check which queries use the most CPU time and disk reads, and decide what to improve on that basis. Sometimes one badly written query in a report run once a day loads the server more than all the other work combined.

First we tidy up indexes and statistics, because this needs no new hardware and no changes to the application. We discuss a more powerful server or changes to the system itself only if that is not enough.

Before you buy a faster server

Check how long the heaviest operation takes after rebuilding indexes and updating statistics. That result tells you whether new hardware is needed at all.

Bundle of optical fibres in coloured light

Test restore of a backup

The backup job succeeds every night and the file is the right size. That still does not prove that a working system can be restored from it.

That is why we restore the backup in a separate environment and measure how long it takes to get back to work. That time decides whether a failure stops the company for an hour or for two days.

We also agree with the company how much data it can lose in the worst case and how long production can be down. These answers decide whether a daily backup is frequent enough or whether transaction log backups need to be added, for example every 15 minutes.

Access and security

Typical risks look like this: everyone works on an account with administrator rights, the password is shared and has not changed since deployment, and the database server is reachable from the internet on the default port. Together they create a situation where one opened attachment can end with the data being encrypted.

We sort out permissions according to the principle of least privilege, close outside access or limit it to specific addresses, and switch on backup encryption. We agree every permission change with the person responsible for the system, so that nobody loses access in the middle of their work.

Charts on a monitor at a desk

Database monitoring

Some database failures give warning signs in advance: disk space running out, routine operations taking longer, growing queues, errors in the event log. You can see them if someone is watching.

We set up monitoring that reports these signs and agree who receives the notifications, so that each alert reaches someone who can act on it.

Who we work for

Most often these are manufacturing and trading companies with an inventory system on MS SQL: Subiekt, Insoft, WAPRO or a custom-built industry solution. The database has usually grown for years, and nobody in the company had the time to look after it.

We support servers on site and in the cloud. We work remotely, and for critical failures also outside standard hours, because a night shift or a Saturday stocktake cannot wait until Monday.

Jakub is very professional and has made running my company much easier.
MrDeXu, review on our Google Business Profile

SQL Server versions, licences and limits

MS SQL comes in several editions, and the differences between them usually show after years. The free Express edition limits a single database to 10 GB (in versions up to and including 2022, and to 50 GB from SQL Server 2025) as well as the amount of memory the engine uses. When the database nears the limit, the system starts to slow down or stops saving data, often without a clear message for the user.

At the start we check the server version, how far the database is from the limits and whether moving to a paid edition makes sense. Sometimes it is better to slim down the database and stay on Express, and sometimes a paid licence is more sensible than yet another workaround.

A separate issue is the version of the engine itself. Releases that no longer receive security patches are sometimes still in use today. We first test a migration to a newer version on a copy for compatibility with the ERP system, and only then carry it out in production.

Working together

How we work together, step by step

  1. Health check

    We check versions, sizes, indexes, backups and server load. We change nothing without your agreement.

  2. Securing

    Backups and permissions first, then optimisation.

  3. Optimisation

    Indexes, statistics, maintenance plan and fixes to the heaviest queries.

  4. Care

    Monitoring, regular reviews and support during failures.

Questions

Frequently asked questions

Mostly not. We schedule index rebuilds and maintenance in maintenance windows, usually overnight. If something requires downtime, we let you know in advance and agree a date.

It can be. Express has a database size limit (10 GB up to version 2022, 50 GB from version 2025) and limits the engine's memory. We check how close the database is to these limits and whether it is better to slim it down or move to a paid edition.

Yes, we treat this as the most urgent kind of request. We then ask you not to attempt repairs on the original: the first step is to secure the current state.

It depends on how much data the company can lose and how long production can be down. When many documents are created during the day, we add daytime transaction log backups to the nightly backup. We keep backups away from the production server and encrypt them.

On the number of servers and databases, their size, the state of backups and permissions, and whether you need a one-off repair or ongoing care with monitoring. We prepare a free quote within 24 hours.

Yes, we provide service and support for InsERT and Insoft software. See ERP system support, InsERT and Insoft.

Let's check what state your database is in

Tell us the server version, the database size and how backups are made. We will tell you whether there is anything to worry about, and prepare a free quote for the work within 24 hours.

All cities and services (858 pages) →
Free quote