1 / 10

Ευαγγελία Πιτουρά Τμήμα Πληροφορικής, Πανεπιστήμιο Ιωαννίνων, Ελλάδα dmod.cs.uoi.gr

Τι θα φέρει το Σύννεφο στη Διαχείριση Δεδομένων : Προκλήσεις και Ευκαιρίες Ελληνικό Συμπόσιο Διαχείρισης Δεδομένων 20 10. Ευαγγελία Πιτουρά Τμήμα Πληροφορικής, Πανεπιστήμιο Ιωαννίνων, Ελλάδα http://dmod.cs.uoi.gr.

affrica
Download Presentation

Ευαγγελία Πιτουρά Τμήμα Πληροφορικής, Πανεπιστήμιο Ιωαννίνων, Ελλάδα dmod.cs.uoi.gr

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Τι θα φέρει το Σύννεφοστη Διαχείριση Δεδομένων:Προκλήσεις και ΕυκαιρίεςΕλληνικό Συμπόσιο Διαχείρισης Δεδομένων 2010 Ευαγγελία Πιτουρά Τμήμα Πληροφορικής, Πανεπιστήμιο Ιωαννίνων, Ελλάδα http://dmod.cs.uoi.gr

  2. What is the new research cloud infrastructure brings to datamanagement? Το νέφος για τη διαχείριση 1 • Δύο νεφελώδη ερωτήματα: • Τι είναι το νέφος; • Είναι κάτι νέο; Ομοιότητες με meta-computing, clusters, grid, κλπ • Shift from local servers to data centers hosted by large infrastructure companies • Pay-as-you-go utility computing • ελαστικότητα (όχι περιορισμοί στους πόρους) + κόστος (χρέωση) με βάση τη χρήση • Προβλέψιμη (;) απόδοση • Ευκολία στη χρήση/ανάπτυξη/επένδυση • economy of scale ("χοντρική-αντί-λιανική") • Scale (data, machines, etc) • Started from an industry need for simplicity/easy of development (Amazon, google, yahoo) DMOD Laboratory, University of Ioannina

  3. Το νέφος για τη διαχείριση δεδομένων 2 • IaaS: Infrastructure as a service • PaaS: Platform as a service • SaaS: Software as a Service • Level of abstraction providedto the programmer by the cloud • Where does data management fit in the stack? • Central point: transactional vs analytical data management DMOD Laboratory, University of Ioannina

  4. Το σύννεφο για τη διαχείριση δεδομένων 3 Approach 1 Build a DBMS on the cloud seen as an infrastructure (hardware) Build a traditional relational DBMS on Virtual Machines Transparency/Elasticity (by allocating new resources, etc) Cost Model ($) Complete Re-design/implementation DMOD Laboratory, University of Ioannina

  5. Το νέφος για τη διαχείριση δεδομένων 4 • Approach2 • Build a DBMS on the cloud seen as a platform • "DBMS"-functionality build on top of: • a "cloud"-like storage, (ie, key-value one) + a programming framework (i.e., MapReduce) • Extend the programming model with database (declarative) functionality (eg Pig Latin) • Query processing on top of MapReduce: query optimization, new implementation of physical operators (eg, joins) • Transaction properties (consistency, availability, durability, fault tolerance (for analytical: amount of work to be redone) • View definition/materilization functionality DMOD Laboratory, University of Ioannina

  6. Το νέφος για τη διαχείριση δεδομένων 5 Approach3 Data management as a service As a web service? API? Models: Key-value stores with provenance (time dimension) Analytical functionality - OLAP style processing DMOD Laboratory, University of Ioannina

  7. Το νέφος για τη διαχείριση δεδομένων 6 Approach4 Use data management "favorites" across approaches Improve storage layer of any cloud Example: Indexes/data partitioning for MapReduce Replication/caching (+materialized views) DMOD Laboratory, University of Ioannina

  8. Mobile and distributed data management is especially relevant. Whatis the "cloud face" of the state of the art there? • There is distribibution at the physical layer (more than one data centers, users geographically distributed) • it costs to move data • transaction management is expensive • Mobility of users • location information • unpredictability DMOD Laboratory, University of Ioannina

  9. How does "old" traditional research change? General Issues • Who owns the data • Analytics for performance • Quality of data/service • Economic Model • Elasticity is central • Scale (amount of data/users) DMOD Laboratory, University of Ioannina

  10. Ευχαριστώ DMOD Laboratory, University of Ioannina

More Related