Guest author contribution
by Jörg Kremer
mip Management Informationspartner GmbH
Head of Consulting / Delivery Manager
The migration of analytical systems to the cloud has long since ceased to be a marginal issue and is now at the heart of strategic IT decisions. Elasticity, scalability and greater agility are promised - all factors that are crucial for data-driven companies. In practice, however, it quickly becomes clear that not every system develops its full potential in the cloud. Some applications benefit enormously from a migration, while others lose their efficiency and stability or are not even viable from a regulatory perspective.
For IT architects who are responsible for infrastructures for data analytics systems, this raises the key question:
Which systems belong in the cloud - and which are better left on-premises?
and which do not | ISR
Things become much more complex with performance-critical applications, such as highly optimized databases or specialized ERP systems. Such systems are often adapted to a specific on-premises environment down to the last detail.
An ill-considered cloud migration can therefore harbor several risks:
- Performance losses because hardware-optimized structures in the cloud cannot be reproduced 1:1.
- Limited configuration options that prevent optimizations such as buffer pool adjustments or storage strategies from being adopted.
- Higher costs if inefficient systems suddenly require more resources.
The lesson learned: extensive testing is essential before a migration. Only pilot projects and detailed benchmark analyses make it possible to assess whether performance and stability also meet the requirements in the cloud.
Data warehouse databases: Why the cloud often sets limits
Existing data warehouse databases are a particularly critical case. Systems such as IBM Db2 or SAP HANA have often been optimized for on-premises environments over many years.
The challenge here is not the technology itself, but the problem of losing configurations. Cloud database services often only allow limited settings, which means that painstakingly developed performance tuning measures are lost.
A practical example: a highly optimized Db2 database runs smoothly on-premises, but loses massive amounts of speed in the cloud - simply because central optimizations cannot be replicated there. For IT architects, this means that migration is risky without very precise analysis and, if necessary, redesign.
Legacy systems: stumbling blocks from the past
Legacy systems are also rarely cloud-friendly. They are often based on outdated operating systems or middleware components that are simply not compatible with cloud platforms.
Even more serious is the fact that these systems are in many cases precisely optimized for specific hardware. The move to a standardized cloud environment therefore almost inevitably leads to performance losses. This can usually only be remedied by extensive adaptations - an effort that often results in a questionable cost-benefit ratio.
In-house developments: When individuality becomes a problem
Highly optimized in-house developments are a double-edged sword. On-premises, they enable fine-tuned solutions with full control over system parameters. In the cloud, on the other hand, greater standardization and limited intervention options apply.
On a day-to-day basis, this can mean less flexibility, reduced performance and limited scalability. Anyone considering moving such in-house developments must weigh up whether the benefits of the cloud outweigh the loss of control and adaptability.
Regulatory sensitive applications: Compliance is crucial
Another field that requires particular attention from IT architects is applications with highly sensitive or regulatory data - for example in the healthcare or financial sectors.
There are two core risks here:
- Data protection and compliance - compliance with GDPR or industry-specific standards must be guaranteed in the cloud without a doubt.
- Loss of control - companies must trust that cloud providers meet all security and compliance requirements.
The migration process for such systems therefore requires not only technical analyses, but also comprehensive risk assessments and approval processes by internal data protection officers.
Practical examples: What we can learn from real projects
Theory and practice often diverge - which makes concrete experience all the more important:
- Retail success story: An e-commerce platform successfully migrated to the cloud and was able to cope with peak loads such as Black Friday without any downtime. Result: higher customer satisfaction and lower costs.
- Critical example of database migration: An industrial company experienced massive performance losses during a Db2 migration to the cloud. The system only stabilized after expensive subsequent optimizations - a classic example of inadequate pre-testing.
- Financial sector: A bank used a multi-cloud strategy to meet regulatory requirements more efficiently. The result: optimized compliance processes and lower operating costs.
- Failed project: A large service provider failed due to a lack of change management. Resistance within the team led to shadow IT, compliance risks and high follow-up costs.
Successful cloud transformations are the result of thorough planning, detailed technical work and active change management.
Jörg Kremer
Conclusion: Cloud yes - but with a sense of proportion
The cloud is not a panacea, but a tool. Cloud-native applications and scalable web platforms are clear candidates for migration. Highly optimized databases, legacy systems or applications with sensitive data, on the other hand, require intensive preliminary analyses - and are better left on-premises or in hybrid architectures in case of doubt.
The most important recommendation for IT architects in the field of data analytics is therefore to check, test and weigh things up before every migration - and never rely solely on the supposed simplicity of cloud services.
Learn more?
mip-Whitepaper
Migration to the cloud
In the white paper from our partner mip, you can find out how to transform your analytics architecture into the cloud in a targeted manner - with a view to performance, security and cost-effectiveness.
Click here for the other parts of the blog article series:
Technical risks in the cloud migration of analytics infrastructures
Guest author contribution by Jörg Kremermip Management Informationspartner GmbH Head of Consulting / Delivery Manager Cloud migration: full ...
What data analytics architects need to consider when migrating their infrastructure to the cloud
Guest author contribution by Jörg Kremer mip Management Informationspartner GmbH Head of Consulting / Delivery Manager Guest author contribution ...
Do you have any questions? Please feel free to contact us!
Wilhelm Hardering
Senior Executive Manager
