Issued by Iskratel; All rights reserved Obr.: 70-121d Analysis of structuring call and non-call services on trial RCS implementation RCS, VoLTE & beyond Worksop, 10-11 October 2012 Valerij Grašič, Iskratel St Structuring t i the th application li ti layer l Lessons obtained AS1 AS2 AS3 HSS UE N UE-N S/P/I-CSCF Issued by Iskra atel; All rights reserved UE-1 UE-3 UE-2 Service placement motivation: to place 5 services to 3 AS Analysis of structuring application layer Many service placement possibilities Starting point for structuring: to group the services to deploy user base on AS set Services Characteristics of services Influence I fl off each h service i Traffic and configuration model framework signaling traffic on S-CSCF and AS Parameters to describe, compare, evaluate configuration user habits More research still need to be done 1 Issued by Iskra atel; All rights reserved RCS iimplementation l t ti options ti on the th market k t Same AS for more local networks More AS for local network RCS AS as part of local network 2 Analysed services on trial RCS implementation Test bed Issued by Iskra atel; All rights reserved UE: Mercuro, Boghe, sipp CSCF, PS AS (Mobicents), AS, HSS HP Compaq Intel Core 2 CPU 6300 Linux i686 (Linux carrier Grade 4.0) 4 0) 1.6 GHZ, Pentium 4, 1 GB RAM CSCF: Iskratel C-IMS Analysed services Basic B i call ll (BC) Use of service OIP/TIP “Basic” signals, PRACK CDIV (Communication Diversion) Service CFU (Unconditional) CAT (Customized Alerting Tones) IM (Instant messaging) Page Mode (use of MESSAGE) Presence Analysed y also Registration Subscribing 3 Application layer arhitecture Issued by Iskra atel; All rights reserved Application servers on IOT 4 Service placement: deploy of services and user base Issued by Iskra atel; All rights reserved Used configurations (K) Grouping of services We defined these grouping rules: Two opposite Services on each AS : placements: all services on each AS (K1) max (K1) vs. part of all services on each AS min (K2 (K2, K3) (K2, K3, K4) number of User base on each AS: services all users on each AS (K2, K3) on AS part of all users on each AS (K1 (K1, K4) Service groups and subsgroups main groups: TAS (MMTel), RCS g p TAS basic,, TAS plus p TAS subgroups: RCS subgroups: RCS IM, RCS Presence 5 Service placement: triggering Triggering iFC (Initial Filter Criteria) for each K One, two or three iFC for each AS Three or four (K2) iFC for each user g ) K0: as TDM ((all services in switch exchange) Issued by Iskra atel; All rights reserved Triggering is made on S-CSCF iFC defines AS where to send messages and where services are executed iFC are stored in HSS 6 Analysed services: call based (TAS/MMTel) Basic call with i h PRACK Call based (TAS) services: Basic call Issued by Iskra atel; All rights reserved Basic call (for K0): Options: “basic”, PRACK CDIV Additional 181 (Call Forwarded) from AS to S-CSCF CAT Additional INVITE from AS to S-CSCF and MRF (Media Resource Function) 7 Analysed services: non-call based (RCS) Presence Non-call ((RCS)) based services: Issued by Iskra atel; All rights reserved IM (p (page g mode)) IM (Instant messaging) page mode Messages: MESSAGE, 200 Presence P (presentity): publish state changes W (watcher): receive state changes Messages from P: PUBLISH, PUBLISH 200 Messages to W: NOTIFY, 200 8 Analysed services: non–call (registration,subscribe) Subscribe: Issued by Iskra atel; All rights reserved We analysed subscribe for presence SUBSCRIBE with “Event: presence.winfo” send for own user SUBSCRIBE with “Event: presence” send for each user in Contacts Registration: Each user must make registration Data are stored in HSS 9 Analysed services: placement to AS set Placement of services Placement of services to AS set: Issued by Iskra atel; All rights reserved Each service can be executed on S-CSCF (with AS functionality), on AS for user A side, and on AS for user B side For example OIP/TIP is executed on AS(A) for A side, then on AS (B) for B side Registration and subscribe are executed y on one AS only CAT and CDIV are executed only on user B side BC and IM can be executed on AS on user A side and on user B side 10 Analysed services: used signals Used signals for services: Issued by Iskra atel; All rights reserved We analysed messages used for services Messages were analysed on user (UE), SCSCF HSS, CSCF, HSS AS and d PS (Presence (P server)) AS (outside AS) We defined messages, size (bytes) and direction Call based services have same messages Size of messages has a wide range 11 A l Analysed d services: i signals i l for f service i placement l t Key findings of the signal analysis of service placement on S-CSCF: Issued by Iskra atel; All rights reserved Traffic for BC+ is almost more than once bigger as traffic for basic call with no additional dditi l options ti Registration and subscribing can have greater impact to traffic than basic call IM in case of two AS has same traffic impact as BC in case of S-CSCF Traffic due presence depends on number of W Traffic for call based services (CDIV (CDIV, CAT) has similar impact as the traffic for BC Signals for service placement Legend for users: G: group of services (number of iFC) C: number of contacts A number A: b off AS user iis registered i t d on BC: number of signals for BC BC+: BC with PRACK and 183 AS1: can be AS for side A or side B 12 A l Analysed d services: i use case for f K1 Issued by Iskra atel; All rights reserved Signals for configuration K1 Use case fo K1: For K1: all services are on each AS For K1: on each AS there is 1/3 of users AS(A) and AS(B) together form AS Traffic is given as number of signals due sevices on S-CSCF and AS (for side A and side B) Reg, subs: given per user IM, BC: executed for both side A and B CDIV, CAT: executed only on side of user B Presence: executed only on one AS 13 E i ti parameters Existing t and d benchmarks b h k Exsisting parameters and benchmarks: Issued by Iskra atel; All rights reserved RFC 6076: 6076 Basic B i Telephony T l h SIP End-ToE dT End Performance Metrics Defined time parameters p Defined also ratio,, attemps RRD: registration delay (REG-200) SRD: session request (INVITE-180) SDD: session disconnect (BYE-200) SDT: session duration (200-BYE) ETSI TS 186 008: IMS/NGN Performance Benchmark Benchmarking of core IMS, no AS Predefined (fixed) set of services No other parameters or benchmarks defined yet IMS/NGN Bencharking g environment 14 P Proposed d parameters t and d benchmarks b h k Proposed time parameters : Issued by Iskra atel; All rights reserved Time parameters (defined): RRD, SRD, SDD, SDT Time parameter (defined, to be expanded) SRDmax: we see that in K3 SRD for users of CDIV and CAT is greater than for users with OIP/TIP only, this fact need to be addressed The Th same is i valid lid ffor SDD Time parameters (used, not defined, to be agreed): g (PUBLISH-200) ( ) Publishing Subscribe (SUBSCRIBE-200) Messaging (MESSAGE-200) Proposed traffic parameters: Traffic parameters Traffic on S-CSCF Traffic on AS Traffic is defined as number of signals SIP transaction (e.g. INVITE-180, BYE200) and SIP dialog (SIP session) are not comparable and can not easly be be combined with non-call services 15 P Proposed d framework f k for f service i placement l t Proposed framework for the service placement: Issued by Iskra atel; All rights reserved Based on the analysis we prepared the framework for the service placement Data for services Which services are used Signals and parameters for services (C (C, P P, W W, G G, A) How the service placement is made Parameters for the configuration g on S-CSCF and AS Traffic: signals Time: session, reg, subs, publish Other: completion ratio,.. Data for users: habits, use of services More detailed research still need to be done for service placement Our plan is to male more tests and simulations for given configurations 16 Open issues for service placement framework Questions detected at presented analysis: Issued by Iskra atel; All rights reserved We have found many interesting completions completions, but some issues are still opened Lack of standardised and agreed parameters and benchmarks time parameters p traffic parameters other benchmarks What parameters to use for traffic comparision? based on BHCA (as in TDM)? based on signals for S-CSCF or AS? based for each service separate (due dedicated AS)? Which metrics and benchmarks still to use for service placement and mutual evaluation? Since service silo is more and more actual (with multitude of different services based on SIP and IMS) answers to these questions are becoming more and more interesting 17