Cloud Managed Services for Google Cloud (GCP)
Your teams innovate. We manage the platform.
Our cloud managed services take on the unglamorous, continuous work of running a data platform: watching it, fixing it, keeping the bill sensible and the access tight, so the people you hired to build things can spend their time building.
Schedule an introduction callSee what we deliver
- 20+ years on Google Cloud
- Google Cloud Premier Partner
- 50+ certifications
- ISO 27001:2022
What happens when nobody owns it
Platforms rarely fail suddenly. They degrade, in ways that are individually small and collectively expensive, in the gap between the project that built them and the team that inherited them.
Incidents are handled by whoever is free
There is no rota, so resolution time depends on who happens to be available and how much of the platform they have seen before. The same incident gets diagnosed from scratch more than once.
The bill grows and nobody can say why
A process runs more often than it needs to, or scans more data than it should, and the cost of that decision compounds quietly for months before it is large enough to notice.
Access was granted and never reviewed
Permissions accumulate as people join projects and rarely shrink when they leave them. By the time anyone audits it, the principle of least privilege has become aspirational.
From firefighting to routine
Most of what makes a platform dependable is boring and repeatable: known owners, a ticket queue, monitoring on freshness and completeness, and someone whose job it is to look at the bill.
Before
- Incidents resolved by whoever is available
- Problems discovered by a user noticing a wrong number
- Cloud spend rising without an owner or an explanation
- Access granted case by case and never reviewed
- Backups assumed rather than tested
After
- A ticket queue, a rota, and resolution times you can see
- Monitoring on freshness, completeness and accuracy
- Billing spikes traced to the process that caused them
- Least-privilege access, reviewed on a schedule
- Backup and recovery planned, documented and exercised
Cloud Managed Services for GCP, run daily
There is a difference between having built something on Google Cloud and having run it for years afterwards. Most of what we know about cost, failure modes and access sprawl came from the second, not the first.
Serverless architecture helps here more than it does at build time: when a workload costs nothing while idle, the optimisation conversation becomes about how often something runs rather than about how much capacity to reserve.
20+
Years building on Google Cloud, since before most of it had a name
50+
Google Cloud certifications across the team
Premier
Google Cloud Premier Partner, with specialisations in data and infrastructure
What Cloud Managed Services cover
Four areas, run as a standing service rather than a project. Weekly standups and a monthly service review are part of it, because the useful conversations tend to be the recurring ones.
Monitoring & support
Day-to-day operation of the platform: watching it, fixing what breaks, and answering the questions that would otherwise sit in someone's inbox.
Proactive monitoring
Continuous oversight so issues are identified and addressed early, which is usually the difference between a fix and an incident.
Incident management
A ticket queue with owners and history, so the same problem is not diagnosed from scratch a second time.
Functional & technical support
Specialists who know your environment handling both the platform and the applications on top of it.
Standups & service reviews
Weekly on the operational detail, monthly on the pattern, with documentation kept current rather than reconstructed.
FinOps: cost management
Someone paying attention to the bill, continuously, with enough detail to say which environment, which product and which process is responsible for a change.
Spike detection
Billing tracked by environment, Google Cloud product and the process behind it, so an increase has a name attached within days.
Query & process optimisation
Batching, partitioning and selecting only what is needed, which is where most large savings actually come from.
Architecture for cost
Serverless where it fits, so idle workloads cost nothing and the bill follows what the business actually did.
DevOps & SRE: observability
Three operational measures applied to the data itself, because a platform that is technically up while delivering yesterday's numbers is not actually available.
Freshness
Whether the data arrived when it was supposed to, checked rather than assumed.
Completeness
Whether all of it arrived, which is the failure mode that produces plausible but wrong reporting.
Accuracy
Whether what arrived is correct, validated against expectations agreed with the business.
Backup & recovery
Regular backups with a recovery plan that has been read by someone other than its author.
Keeping access tight as the organisation changes around the platform, which is mostly a matter of reviewing it on a schedule rather than when something goes wrong.
Identity & access management
Least privilege applied and maintained, so permissions shrink when people move on as well as growing when they arrive.
Access reviews
Scheduled rather than triggered, which is the only version that catches the slow accumulation.
Compliance support
Evidence available when an auditor asks, because it was being collected before they did.
Two things worth measuring
One a number, one a sentence from the person who has to live with the platform.
eCommerce · cost optimisation
BigQuery costs cut by more than three quarters
An eCommerce client had inherited processes from another vendor that updated hourly, retrieved more data than they needed, and got worse as the tables grew. We consolidated row-by-row updates into batches with a caching layer, rewrote the queries to select only the fields in use, and partitioned the tables. Data scanned went from gigabytes to megabytes: 38% lower within 20 days, and 77% lower after a month.
Read the case study →
Fritjof Haalboom · Head of Technology, Body&Fit
Continuity for an eCommerce platform that cannot stop
Body&Fit needed faster incident resolution and better visibility into how their data platform was performing and what it cost. We run daily monitoring and bug fixes, incident management through JIRA, weekly standups and monthly service reviews, alongside data quality assurance and access management.
"The managed services Crystalloids delivers is giving me the reassurance that the platform and the services integrated on the Google Cloud Platform are performant and operating as it should."
Read the case study →
Before you hand over the platform
Do you have to have built it to be able to run it?
No, and a fair amount of what we run was built by somebody else. It does mean the first few weeks involve more reading than fixing, and we would rather be honest that an inherited platform takes a little time to understand than pretend otherwise.
What does this cost against hiring someone?
It depends mostly on how much platform there is. The argument for a managed service is usually less about the hourly comparison and more about coverage: one internal specialist is a single point of failure who also takes holidays.
Where does your responsibility end and ours begin?
Agreed at the start and written down, because ambiguity here is what produces the incident nobody picks up. In practice we tend to take the platform and the pipelines, while product decisions and business logic stay with your team.
Can we take it back in-house later?
Yes, and it is worth designing for even if you do not intend to. Everything runs in your own Google Cloud environment with documentation your team can read, so leaving is a decision rather than a project.
How is access to our environment handled?
Least privilege, reviewed on a schedule, with our own access subject to the same rules we apply to yours. We are ISO 27001:2022 certified, and for regulated clients we work to whatever additional framework applies.
Next step
Want the platform to stop being the thing that interrupts?
If incidents are landing on whoever is free, or the cloud bill has grown without an explanation, those are both good places to start a conversation.