Jinkō platform known limitations

updated 8 mths ago

Performance and scalability by design

Jinkō is designed to be a fast, robust, and scalable solution, aimed at revolutionizing how simulations in drug development are conducted. Just like how a Formula 1 car, built for speed and agility on the racetrack, isn't suited to hauling heavy payloads, and a lorry, designed for carrying heavy loads, wouldn't perform well in a high-speed race, Jinkō has its operational limits. These limits, akin to the soft boundaries on a racetrack, guide users on how best to utilize the platform to achieve optimal performance without compromising the overall system integrity.

While Jinkō encourages exploration beyond these boundaries, it's crucial for users to understand that doing so is at their own risk. Should their simulation demands excessively strain the platform, potentially degrading the experience for others, we reserve the right to intervene by canceling the simulation.

The outlined limits span various Project Item types, from the number of components in a Computational Model to the size and number of Protocol Arms, Trials, Trial Visualizations, Documents, Bibliographic references, and overall Project Items. These guidelines are established to maintain Jinkō's high performance and reliability while accommodating the diverse needs of our users, ensuring that Jinkō remains a cutting-edge platform for simulation in drug development.

The limits

Project Item (PI) type	Limit type	Soft limits	Rationale
Computational Model	Number of components	5,000	Edit is slow for very large (10k+ components) models and graph lags for medium size (1k +) models when depth ≥ 3
Virtual Population	Size = Number of descriptors x Number of patients	1,000,000	Size of uploaded file (stream)
Protocol Arms	Number of arms	100	Trial monitoring needs to be improved both in core and front to account for a large arm set
Protocol Arms	Number of descriptors	1,000	Display in the front (no pagination or lazy loading for now)
Trial	Size = Number of patients x number of arms	100,000	The limits are the trial monitoring and the download of results for now, notably if the user selects too many descriptors
Trial Viz	Time Series Aggregate (num patients)	1,000	The time series aggregate does not work well beyond 1000 patients
Document	Numbers of words	500,000	Limit is more on the client side and can vary from one user set-up to another, this holds for a machine with 16Gb of RAM and 4 CPUs
Document	Number of Cm renderers / equations	200	Limit is more on the client side and can vary from one user set-up to another, this holds for a machine with 16Gb of RAM and 4 CPUs
Bibliographic reference	Scientific article max size	50MB	Rendering / Knowledge processing
All PIs	Max PI count per project	10,000	Maintenance, navigability,Re-use (project carbon copy),Back-up
All PIs	File size upload	2GB	Network, backup & maintenance (hard limit on backend is 15GB because of large json vpops)
All PIs	Max number of users editing the same PI at the same time	10