The hospital was founded in 1999, the Shenzhen Municipal Government investment into health care, teaching, research and prevention as one of the modern three general hospitals. 2017 review by the hospitals. 2018, total hospital emergency door about 2.92 million passengers, nearly 74,000 people discharged patients in surgical volume 49000 units times. The hospital in Shenzhen have high visibility and recognition.
Industry Challenges
With the explosive growth of the hospital's outpatient, radiology and other departments to add a large number of image acquisition devices, PACS storage of large amounts of unstructured data, legacy storage system faces the following challenges:
l insufficient access performance
A large number of image acquisition devices produce video images and need to go through massive parallel processing and storage of PACS, it will lead to existing storage performance face greater pressure and influence doctors to use access to video speed.
l fast data growth, data volume
Image file monthly new number nearly 10 million, an annual growth amounted to hundreds of millions, storage space of up to PB-level and file saving time requirement of not less than 15 years, can not meet the needs of traditional storage.
No data classification and archiving l
Prior to the PACS archive storage device set brand variety, no unified data lifecycle management, storage systems need to be reasonable classification and archiving, depending on performance and capacity to meet compliance requirements.
l image file can not be real-time access to
Prior to image storage system supports access to image data within six months, such as exceeding six months, the clinical practitioner can not timely access to history can not even image files, thereby affecting the efficiency of seeing patients in urgent need of efficient, intelligent storage system to meet business needs .
l need to centrally manage and share information
Prior to PACS can only handle manually copy the good image to the video distribution system, and to provide access to clinical practitioners perform diagnostics, data stored in other sections of the department workstations, and centralized management of information can not be shared.
solution
Based on this, the medical units must be upgraded to traditional storage architectures and data call mode, in order to cope with the current demand for rapid expansion of the store. After several industry research and product comparison test, the hospital finally adopted Angeles rock distributed object data storage solution architecture as shown below:
Transformed into the underlying storage cloud storage pool architecture and original traditional heterogeneous storage, unified management. Image Publishing System directly read image data of the thermal storage pool for nearly a year, and without the production of PACS for image copy, save storage resources, and accelerate the operating efficiency of PACS systems and publishing systems. Other departments (dermatology) documents using seafile unified document management system for centralized management and shared, centralized data storage pool in the storage temperature. Cloud storage pool can set archiving policies directly, without the need for data classification upper system archiving.
Related capacity Design:
Prior to the three hospitals PACS system to produce one day 300 000 images per CT image size is 500K, each DR, chest size is 12M. One day increase the amount of data nearly 200G of the year, an increase of up to 72T.
Cedar Rock PACS imaging systems distributed storage architecture
historical data archiving total production PACS system 150T, other departments (dermatology, gynecology and obstetrics, Neurology and Neurosurgery, etc.) need to 30T storage space for centralized storage and management. Cloud storage space requirements as bare 500T (heat storage capacity of the bare cell 200T, temperature storage pool 250T, optical storage pool 50T), or the underlying data using secure copy protection mechanisms to protect the erasure of data.
Distributed storage scheme of the latest design, the heat cell temperature and cell capacity were designed according to 1 and 2 years to store data, i.e., data is written directly in 19 hot pool, in 17 to 18 years of direct transfer of data to the two Wenchi, 17 years ago to migrate archived data directly to optical storage. Data after 3 years stay disk storage, offline archive to cold storage (optical memory). According to current statistics, the hospital annual increase of about 80TB of image data, in accordance with the storage time, temperature and thermal pools and pool were designed 100TB and 200TB of usable space, the follow-up business growth, expansion only optical storage can be. If the result of business growth, the annual growth rate to improve the data, press the demand and the ratio of hot, warm and cold for expansion.
Calculating a final raw capacity storage device P should consider the available traffic capacity requirements p, hard disk breakage rate r, and copy or erasure code (4 + 1) of the available space s, which is calculated, the hard disk breakage rate r formula: r = (1000/1024) 4 = 0.91
Two copies of the available space s1: s1 = 1/2 = 0.5; erasure code rate available space s1: s2 = 4/5 = 0.8
Raw capacity and available capacity P p, and a hard disk fracture ratio r, the relationship between s erasure code rate available space: P = P R & lt s
Thus raw capacity (in terms of hard disk manufacturer's specifications) can be calculated by the following equation: P = p / (r * s)
According to data volume growth for the year 72T-80T, and taking into account the requirements of the future and increase image clarity of image data, the image need access to the original image. Thermal storage pool free capacity planning PACS system 100T (data stored for one year), due to the high performance requirements of the reader, store a copy of the program is recommended two. And mainly small file read and write frequently, in order to consider the cost, the main storage disk with SATA machinery, the use of two SATA SSD to do some critical metadata storage acceleration.
The available capacity of the heat storage pool is calculated as follows: 200T 0.91 0.5 = 91T
Read and write performance since the storage pool temperature requirements are not high thermal storage pool, storage use erasure codes EC embodiment, additional other sections (dermatology) temperature data stored in unified storage pool, storage server may consider a multi-bay server, reducing storage hardware costs.
Available Space formula: 250T 0.91 0.8 = 182T (data kept for two years), the cold storage tank (optical storage) of free space is 50T.
Program Value
Through this transformation program, with cedar rock mass data storage objects (SandStone MOS) solutions, full use of existing storage devices, PACS will replace the image file storage system into SandStone MOS, production systems store files in a scalable, large capacity storage pool, simplify infrastructure, to facilitate unified management and operation and maintenance monitoring. The transfer of data to a lower-cost hardware; 2 years or more archive data permanently stored on the Blu-ray storage, to meet regulatory requirements to retain more than 15 years. By nanotube legacy storage system, centralized management and seamless data migration. The system uses the index to find alternative hashing access files, merge files using innovative ways of data storage, optimized for small file access performance. SandStone MOS built-in search engine, breaking silos of data, data sharing, greatly enhance the operating efficiency of PACS.