Federated Learning: Reduce AI Costs by Up to 35%

Stuck in data silos and frustrated by costly data movement? Break free with Scalytics Connect, our innovative AI-powered platform. Analyze any data, anywhere, directly at its source – no need for centralizing or copying. Unify your existing data platforms and processing engines for seamless collaboration and unlock faster insights with on-site AI and machine learning. Scalytics Connect not only empowers smarter decision-making but also recovers up to 35% of your data spend through improved management practices. Capture double-digit savings within six months by optimizing your current data stack. Unleash the true power of your data with guaranteed security and privacy – all with Scalytics Connect.

‍

CapEx cost reduction by unifying existing data platforms

Part of a recent customer integration involved comparing performance before and after deploying Scalytics Connect, as opposed to using standalone Spark instances for big data analytics. While research often cites Spark as the fastest big data system, this real-world case study offered valuable insights. Scalytics Connect functioned as a federated data access system, feeding processed data to Spark's machine learning modules. To reflect realistic user workloads, we focused on three main areas:

‍

• text analytics (e.g., word frequency, word synonyms, inverted index creation) 
• data analytics (e.g., aggregate queries and join queries)
• machine learning (SGD, K-Means, and cross-community pagerank)

‍

For this comparison, we considered a single AWS cloud instance of two popular types: m4 ($2.42 / h) and T3 ($8.786 / h). We assume that the user keeps running the instance 8h / day for performing its data analytics. The table below illustrates the benefits benefits in terms of time and monetary cost savings. Remarkably, we observe that using Scalytics Connect always translates to time and cost savings: it allows users to save over $200,000 per year in the above-mentioned setting. 

‍

‍

Reduce OpEx, cut CapEx costs by reusing your current data stack

Based on our experience, organizations may free their employees by one-third by ignoring cost savings for employing IT staff and can immediately redeploy their current workforce. For instance, to keep a Spark cluster of 25 nodes in AWS and run around 5 consulting AI projects, the typical team size overall is 14 team members:

‍

• Backend developer (5) 
• System specialists (2)
• Data scientists / Data analyst (4)
• Project managers (3) 

‍

Implementing Scalytics Connect reduces the average team size to 7 staff members:

‍

• Backend developer (2) 
• System specialists (1)
• Data scientists / Data analyst (2)
• Project managers (2) 

‍

This also leads to significant cost reductions for the entire firm. Due to the prolonged use of previously existing data processing platforms, such as Hadoop or Spark, and his commercial versions, our clients often save 35 - 40% OpEx expenses and on average more than 50% CapEx costs when using Scalytics Connect. Please keep in mind that the OpEx savings may be promptly redeployed to drive more projects at the same time.